# $Id: mobot-survey.text,v 1.3 1996/04/08 01:13:21 wlim Exp $ ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; ;;; ;;; ;;; ;;; RESPONSES TO INFORMAL SURVEY ON DEVELOPMENT ENVIRONMENTS ;;; ;;; FOR MOBILE ROBOTS ;;; ;;; ;;; ;;; ;;; ;;; ;;; ;;; Updated: Tue Dec 6 08:59:45 1994 ;;; ;;; Created: Sat May 23 09:37:24 1992 ;;; ;;; ;;; ;;; Maintained by: wlim@lehman.com (for now) ;;; ;;; ;;; ;;; Please send updates, additions, corrections, etc. to: ;;; ;;; wlim@lehman.com ;;; ;;; ;;; ;;; A complete version of this survey including detailed ;;; ;;; descriptions of the various projects is available via ;;; ;;; anonymous ftp from the host ftp.ai.mit.edu as the file ;;; ;;; /pub/mobot-survey.text. ;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Organization Robot Development Languages & Type HW Environment SW Enviroment ============= ==== ============== ============= Alcatel Alsthom Indoor robot SPARC II, VME proc VxWorks, MOTIF Recherche (AAR) Outdoor robot Brown U. a)Mobile robots SPARC I & II, OS/9, GNU Emacs, Xlib, VME 68030 MOTIF, Forth, C, C++ b)2 RWI B12's Sparc 10s/Solaris offboard UNIX, Motif, C++, Lisp, Rex, X c)2 RWI B24's 486 Linux onboard, arms CMU a) SM^2(walker) VME 68020 & 68030, Chimera II RTOS, C, Sun Sun tools b) AMBLER VME 68020 & 68030, MOTIF, VxWorks, SPARC II, Iris X windows, C c) Mobile Mani- Z8088s, Sun IPC & ELC, X Window, Lisp, C, Hero pulator NeXT Basic d) Mobile robot SPARC, Iris, Mac, X windows, Openwindows, VME, Maspar, Titan VxWorks, Chimera RTOS, TCA, GIL, LISP e) Navlab Sun-4 X windows, C Colorado Sch. Denning MRV-3 Sparc IIs, IPXs, C, X11, Khoros, potential fields of Mines IBM RS/6000s X11 visualization tool (homemade) Colorado St. 6-legged robot 68HC11EVM, AT C Cornell U. 2 mobile robots Gespak 68000, Intel Scheme, Lucid Lisp (robot with 80c196, Sun(?) tank-tread base coming) Cray Research(?)Mobile robot MC68HC16EBV, 386 Assembler Georgia Tech Denning DRV-1 SUN IPC, Decstation, X windows, C, & MRV-II Microvax II Lisp Grumman CRC SmartyCat Mac II's/IIci's, C, CLOS, LISP, SAL (Cybermotion uExplorer, SGI VxWorks(soon) K2A) 68030 VME board(soon) LLV (Grumman SGI, 68030 VME board, C, CLOS, LISP, SAL Long Life Veh., mini-boards. the US Postal Truck) IBM TJ Watson TJ, TJ2 Symbolics, RS/6000, LISP, CLOS, CLIM, (1989?-1992) 286, 386, Suns C, X-windows, MOTIF, GNU Emacs ISX Corp Subsumption Mac II cx's/ci's C(?) JPL 7 robots Suns to 6811 RCCL, ALFA McGill U Mobile robot Sparcs, mc68hc11, PC GNU, X, Small-C C, C++ QUADRIS SUNs, Macs, C-40, 68K C, X-windows, IRIS GL Michigan Tech. Tracy 6502, Apple IIe, SUN, C, Assembly U. 68HC11 Unnamed(Andros) Minirobots 6811 MIT 20 robots Mac II & IIsi, HC6811 Behavior Language GOPHER (ISR R2) 68332, Mac, Sun GCC, Behavior Language, Lisp, X-windows Polly VME, 6811, Mac Senselisp(Scheme) SOZZY(homemade) 6811, Mac Lisp, Behavior Language MITRE Denning MRV-1 MacQuadra, uExplorer Lisp, REX/GAPPS, C, C++ Northeastern U. Lobster Robot HC11, Mac C, Pascal, Assembly Phaeton Sun 4/330, Mac C, epsilon (Cognex), (Denning MRV3) X-windows NRC of Canada EAVE Mac II's, 68020's C, HARMONY OS, MacAPP (Cybermotion) NC State Mobile robot VME 68020 & 68040, OS/9, P/NET Osaka U. Homemade VME 68030, SUN IPX, C, X-windows Sparc 2 Purdue U. PETER Sun4, 68030 C, VxWorks (Cybermotion) SRI FLAKEY Sparc10/30, Z80 Lucid Lisp, C, X-windows Stanford Landmark based Mac IIci C, LISP Navigation (Nomadic) Swiss FIT Mobile robot Mac MacMETH, Modula-2 U of Central a) 6-leg walker Commodore 64 SuperC, C b) 6-leg walker Amiga 500 C U of Edinburgh a) ALDER 8052, SUN, PC Basic (Fischertecknik) b) CAIRNGORM 68000, SUN C (Fischertecknik) c) Bill (RWI) PC, transputers C d) Ben Hope(RWI) transputers C e) (LEGO based) 68000 C, CPL U of Mass., Denning DECstation 5000, C, LISP Amherst Sparcstation U of Michigan BORIS (TRC) 486, Decstations, SGI, Borland C++, FORTH, DOS RS/6000 CARMEL (K2A) 286, 486, (ditto) Borland C++, FORTH, DOS MAVERIC 486, Sparc 10, Lisp, GCC, Borland C++, X, DOS Datacube, (ditto) U of New Underwater Sparcstation, VxWorks, C(?) Hampshire robots CMOS VME boards U of South Cybermotion K2A Z-80, 68000 PASM, GEHPL, UNIX, Carolina & K3A, Heathkit DOS/Windows Hero 1 ET-18 U Wash. Denning HP 9000 series 300's, Gensym G2, OS/9 68000 LLAMA (Forth), Lisp, C Worcester Poly- James NEC 76310, 68HC11, Assembly, Small-C (DOS) technic Inst. (RWI B12) Gateway 2000 PC Wright Lab, Hero 2000 286 MS C (DOS), Assembly Wright-Pat. Air Force Base VTT (Technical Akseli HP-1100, 386 MS-DOS, LynxOS (soon) Research Center C of Finland) Organization Submitted by ============= =========== Alcatel Alsthom Recherche (AAR) noreils@aar.alcatel-alsthom.fr Brown University a) kjb@cs.brown.edu b&c) jgm@cs.brown.edu CMU a) dstewart@IUS4.IUS.CS.CMU.EDU b) Reid_Simmons@FREESIA.LEARNING.CS.CMU.EDU c) Reid_Simmons@FREESIA.LEARNING.CS.CMU.EDU d) nivek@frc2.frc.ri.cmu.edu e) Chuck_Thorpe@IUS4.IUS.CS.CMU.EDU Colorado School of Mines rmurphy@mines.colorado.edu Colorado State University jn163051@longs.lance.colostate.edu Cornell University jar@cs.cornell.edu Cray Research kilian@palm.cray.com Georgia Tech arkin@cc.gatech.edu Grumman CRC wlim@lehman.com IBM TJ Watson nhaas@watson.ibm.com ISX Corp cfriedla@isx.com JPL gat@robotics.Jpl.Nasa.Gov McGill University dudek@mcrcim.mcgill.ca Michigan Technical University pwelliso@major.cs.mtu.edu MIT ian@ai.mit.edu maja@ai.mit.edu masaki@ai.mit.edu MITRE slack@starbase.MITRE.ORG NRC of Canada Liscano@IIT.NRC.CA Northeastern University jdc@meceng.coe.neu.edu North Colorado State University aras@eceris.ece.ncsu.edu Osaka University ishiguro@sys.es.osaka-u.ac.jp Purdue University kak@ecn.purdue.edu SRI konolige@ai.sri.com Stanford eno@leland.stanford.edu Swiss Federal Institute of Technology vestli@ifr.ethz.ch University of Central Florida rbc@engr.ucf.edu University of Edinburgh bridget@aifh.edinburgh.ac.uk ulrich@aifh.edinburgh.ac.uk University of Massachusetts, Amherst connolly%rabbit@cs.umass.edu University of Michigan koss@engin.umich.edu pkenny@eecs.umich.edu University of New Hampshire rg@nymph.msel.unh.edu University of South Carolina gedney@ece.scarolina.edu University of Washington bob@robocop.ee.washington.edu Worcester Polytechnic Institute tyson@WPI.EDU Wright Lab, Wright-Patterson AFB bartha@corsair.aa.wpafb.af.mil VTT (Machine Automation Lab., vjs@kau.vtt.fi Tech. Res. Ctr. of Findland) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; ;;; ;;; ;;; ;;; Edited (minor) messages (detailed responses) ;;; ;;; ;;; ;;; ;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Date: Thu, 23 Apr 92 11:41:43 PDT From: cfriedla@isx.com Our group works with and builds Brooks Subsumption Robots So we all use Mac II CX or CI . *********************************************************** * * * Carl Friedlander Ph.D. (818)706-2020 * * ISX Corporation FAX (818)706-2056 * * 4353 Park Terrace Drive Page (818)592-5410 * * Westlake Village, CA 91361 * * * *********************************************************** Date: Thu, 23 Apr 92 12:07:31 PDT From: eno@leland.stanford.edu I'm working on a project on landmark based navigation, we use: 0) A robot from Nomadic Tech. (based here in Palo Alto) which has three rings of sensors i) 16 sonars ii) 16 infrared sensors iii) 16 fairly useless bumpers.. the only processing done on board is control of motors and sense cycles and communication via the radio-modem 1) I'm using a mac IIci because basically that's what was available and I think until recently, that was the only platform for which they'd written an interface for. Until now, processing power hasn't really been an issue. 2) The interface and simulator are both written to run in Allegro Common Lisp, which is nice. The on board stuff for the robot is, as far as I know (pardon my ignorance, but I haven't had to mess with it yet), C for the 6811s which written, compiled, and downloaded from any old PC with the proper connections. hope this helps, If you want, I can get more technical specs concerning on board processing, sonar and IR, etc.... Ben Date: Fri, 24 Apr 92 10:58:33 +0200 From: noreils@aar.alcatel-alsthom.fr ( Fabrice Noreils ) Dear Willie: We have two robots at Alcatel Alsthom Recherche (AAR): The first one is a robot for indoor environment and has two motorized wheels and is equipped with a ring of 16 ultrasonic sensors and laser stripe range finder. This work focusses on world modeling (using sonars), navigation (obstacle avoidance and path planning), and mission planning. The second one is a robot for outdoor environment and has four tiltable tracks robot equipped with force sensors, an inertial reference system, and a laser-stripe range finder. This robot is either teleoperated or autonomous and thus it is able to model the terrain (with the laser-stripe) and plan a trajectory (by modifying the orientation of the tracks). Both Robots software run on a VME multi-processor hardware (up to 10 boards) supporting a real-time operating system (VxWorks). Softwares are first developed on SPARC II workstations connected to Ethernet ( workstations run X windows and MOTIF) and are then downloaded into robots. If you need more information, please do not hesitate to contact me Cheers - fabrice ------------------------------------------------ Fabrice R. Noreils Research Scientist Alcatel Alsthom Recherche Route de Nozay 91460 Marcoussis - France email: noreils@aar.alcatel-alsthom.fr phone: 19 - 33 - 1 64 49 17 73 FAX : 19 - 33 - 1 64 49 06 95 ------------------------------------------------ Date: Fri, 24 Apr 92 19:20:39 MDT From: jn163051@longs.lance.colostate.edu Six legged walking robot, legs in stretched hexagon, each leg has three DOF, legs activated by pneumatic cylinders, no intermediate positions extend/retract only, eight walking directions, spin CW/CCW, limping mode involves walking on five legs only, speed 0.4 m/s, IR sensor detects range and direction to reflector obstacle. Controlled by on board motorola 68HC11EVM, code generated in C on an AT. Regards -Joel Date: Sat, 25 Apr 92 11:47:40 CDT From: kilian@palm.cray.com (Alan Kilian) I have a mobile robot with a rotating ultrasonic sensor. It has a Motorola MC68HC16EBV Evaluation board mounted on it. I use the Motorola Assembler and evb16 debugger on a Zeos Notebook 386 PC to develop the software. -Alan Kilian kilian@cray.com 612.683.5499 (Work) 612.729.1652 (Home) Cray Research, Inc. 655 F Lone Oak Drive, Eagan MN, 55121 Date: Sun, 26 Apr 92 12:49:00 +0200 From: Sjur Jonas Vestli [stuff deleted --- wlim] Own design, documentation available on request. [stuff deleted --- wlim] Macintosh, connected to robot via serial link (wireless lan in the future when frequencies are released here), connected to setral facilities via Appletalk. [stuff deleted --- wlim] Own development (on both Mac and Robot). Both based on MacMETH (Mac Modula-2 ETH). Own real time kernel for the robot. [stuff deleted --- wlim] Modula-2 Sjur J. Vestli Institut fuer Robotik Swiss Federal Institute of Technology. Date: Sun, 26 Apr 92 12:43:29 -0400 From: kjb@cs.brown.edu (Ken Basye) 0) We have a couple of robots built on RWI bases with Denning sonar modules; the on-board computation is done on New Micros Inc. 68HC11-based single-board computers, which have a FORTH kernal in ROM. 1) We use SUN sparc 1's and 2's networked with ethernet (TCP/IP) 2) We edit with GNU emacs, write graphical stuff with Xlib and Motif. 3) We use FORTH on the robot, and C++ on the workstations. ----------- We have another robot under development - here's my best guesses about that: 0) The robot is based on a much larger RWI base (about 26 in. diameter, 1200 W/hrs of battery power). The on-board computer will be a Moto. 147 - a 68030-based VME board with lots of interface, including ethernet. This will be running OS-9 with realtime extensions. 1) same workstations 2) Same environment on workstations, OS-9 environment on robot 3) We'll use C or C++ on the robot instead of FORTH Date: Mon, 27 Apr 92 07:32:25 -0400 From: arkin@cc.gatech.edu (Ronald Arkin) All I have time for: 1) Denning mobile vehicles - DRV-1, MRV-II. 2) Sparcstation IPC, Decstation 5000/120, Microvax II, ethernet links 3) Standard unix environments and tools (X windows) 4) 90% C, 10%lisp. Date: Mon, 27 Apr 92 8:41:18 EDT From: dstewart@IUS4.IUS.CS.CMU.EDU [stuff deleted --- wlim] Self-Mobile Space Manipulator (SM^2). (Principle Investigators: Ben Brown (hbb@ri.cmu.edu) and Takeo Kanade (tk@ri.cmu.edu)) This walking robot has been designed for space station trusses, to both travel on it, and manipulate objects (e.g. other parts of the truss structure, satellites, or tools required by astronauts.) [stuff deleted --- wlim] VMEbus-based system, with several Ironics MC68020 and MC68030 single board computers (SBC). Each SBC is running the Chimera II Real-Time Operating System (finger 'chimera@cmu.edu' for more info on it). The system also contains ADCs, DACs, and PIOs for I/O. The VME system is hosted by a Sun workstation, on which development is performed, using standard Sun tools (e.g. cc, dbx, X or Sunview, etc.) and the Chimera II support tools. Communication between the Sun and VME system is through BIT3 VME-to-VME adaptor. The robot is tethered, hence does not carry the computer on board. Multiple CPUs are used to get additional performance. All interprocessor communication is over the VME backplane and processor transparent, and the capabilities are provided by Chimera II. [stuff deleted --- wlim] C for everything. We also use "reconfigurable systems software" methodology for developing modular and reusable code (finger 'chimera@cmu.edu' for references). Dave Stewart Date: Mon, 27 Apr 92 16:12:33 -0400 From: aras@eceris.ece.ncsu.edu (Caglan M. Aras) [stuff deleted --- wlim] We are using a 6 processor VME system (4- 68040, 2-68020) with OS/9 and P/NET, a distributed pipe manager for multiprocessing. We also have a datacube frame grabber and framestore. We use the VME both for development and for the actual control. In addition we have used the Image program on the macintosh to process some of the images before writing specific programs on the VME system. -- Caglan M. Aras | Robotics and Intel. Systems Lab. RISL- Box 7911 | aras@eceris.ece.ncsu.edu North Carolina State University | ph: 919-515-5405 Raleigh, NC 27695 | Date: Tue, 28 Apr 92 20:33:24 EDT From: nhaas@watson.ibm.com [stuff deleted --- wlim] The short answer is that we use about 70% Symbolics, 15% RS/6000; 5% DOS machines of various flavors, and 10% Suns. Over time, the RS6000s will increase. We have (as you know) 4 RWI 12" bases, and the Esched Scorbot III arm. We have since acquired a Zebra Zero Gravity arm, specially lengthened to our specifications. We use Lisp, except for Alberto, and the vision people, who use C. Jon has lately been trying to use Microsoft Small C on one of the DOS machines. He doesn't like its lack of debugger, or lack of decent debugger. On the RS6000, we use Lucid lisp, with CLOS and CLIM. It works OK for me, but Jon does a lot of array-consing, and the garbage collector seems to have a bug where it runs out of space eventually. (I don't see this in my work, though.) We are still running the beta-test version they sent us last summer; they say there'll be a new, real production version of Lucid coming in a few weeks. The debugger for this Lisp is OK but not great by any means. There is one guy around here who has Allegro; he says the debugger is really nice, but still only about 75% of what Genera provides. CLOS seems to have no problems of any kind. I like CLIM a lot, especially the colors (!) and 8-bit gray scale (Symbolics' displays have to dither to show gray), but we still have version 1.0, which I think has some flaky (repeatable, but stupid) behavior which I expect will be ironed out in the next release of CLIM (1.1? 2.0?). We of course use X-windows for everything on the AIX or UNIX machines. Mostly use Motif (mwm); Alberto uses twm, if I remember correctly. The DOS machines mostly do NOT run windows, but there are 2 here that do. (We have 286 and 386 processors; no 486's so far. All stationary. On board the robots, only Jon's 68HC11 network, coupled by radio (ARLAN is the manu- facturer) modems to the Symbolicses. Networking: Symbolics' Chaosnet, Ethernet, and PC token-ring LAN. TCP/IP, NFS-mounting between all machines except PCs. We do not have VxWorks, OS-9, CONDOR, or anything like that. The C people on the RS6000 use something called SabreC, which they claim is great in all respects. Alberto uses Khoros for image-munging, and another tool with a name like XRS. He likes them. For figure document preparation, there's I-DRAW, which might be available on both UNIX and AIX, which creates just wonderful color foils, and I gather it's not hard to run. Hope this will do. Feel free to ask questions on anything I've forgotten to address. -Norm Date: Wed, 29 Apr 92 11:38:10 EDT From: rbc@engr.ucf.edu (Robotics Club) [stuff deleted --- wlim] 6 legged walking machine, designed for the SAE Walking Machine Decathlon. Each joint powered by a DC Brush motor, with feedback being given by potentiometer. Digital sensors include foot on ground, obtacle in leg path, and platform level. [stuff deleted --- wlim] uh, a commodore 64. They are cheap, and we have had experience with them. We use a custom made memory map controller using four 6522 Versatile Interface Adapter chips, and one ADC8017 16 channel, 8 bit A/D. [stuff deleted --- wlim] ... We use SuperC, a C language compiler for the C64. [stuff deleted --- wlim] C Note. We are moving to another robot, same configuration, we will use a 68000 based Amiga 500, another custom controller card with four 6522's and 2 ADC8017's, development environment is DICE, Dillon's Integrated C Compiler, a Freeware C compiler. We use amiga's and CBM equipment due the the fact that interfacing to the 68000 is very simple, and we use them as our own computers. The Amiga has a very nice near-real time multitasking operating system, which will run in 512K. WITH your code. Date: Wed, 29 Apr 92 12:27:49 EDT From: nhaas@watson.ibm.com I meant to mention in my previous note that the typical way to run Lisp under AIX is to run it in a GnuEmacs buffer. With Lucid, there is no partic- ular synergy; with Allegro, there is the ILISP package (which I think was written at CMU), which is a suite of GnuEmacs-lisp routines that tightly couple the editor to the lisp, so that you can do "meta-point", for instance, and many more good things. I implemented a couple of these sorts of things myself to work with Lucid, but didn't achieve the same level of performance, because I was just passing text strings to *standard-input*, and Lisp read them whenever it felt like it. The ILISP system uses a TCPIP "socket", and starts a Lisp process which is dedicated to reading it; this process is asynchronous with the main lisp process or processes. Date: Tue, 05 May 92 11:11:12 -0400 From: Reid Simmons Willie, Sorry for the delay in responding. I'll answer regarding my two robot projects: the Ambler Planetary Rover and the Mobile Manipulator. If you'd like, I could forward this on to other projects at CMU (NavLab, HMMWV, Tessalator). Reid I. AMBLER: 0) A brief description of your robot configuration. Ambler is a six-legged robot designed for autonomously walking across rugged terrain, such as that found on Mars. The Ambler features orthogonal legs (RPP mechanisms) that decouple horizontal and vertical motions. It has a novel *circulating gait*, in which trailing legs recover through a central body, past the other legs, to become the new leading leg. The Ambler uses a scanning laser rangefinder to provide 3D terrain maps, and six-axis force sensors in each foot to indicate terrain contact. The Ambler also has incremental and absolute encoders on each joint, and two inclinometers to indicate tilt and roll of the machine. 1) Type of workstations (e.g. SUN, Silicon Graphics, IBM RS6000, HP, IBM PC/XT/AT/286/386/486, Macintosh, ...) used and why. If you more than one workstation, please say how they are networked (e.g. ethernet with TCP/IP, Appletalk,...) The Ambler uses 9 Creonics motion control boards for low-level servo control. The real-time controller is implemented using two Motorola CPU boards that communicate via a VME backplane. Perception and planning software run on two on-board Sun SPARC II workstations, which communicate with each other and the controller via thin-wire Ethernet and TCP/IP. The perception software communicates with the laser scanner via an S-bus to VME-bus converter. The various hardware choices were made mainly for reasons of availability and compatibility with our existing development environment. We also make use of two simulators -- a 2D X-based simulator, that runs on Sun workstations, and a 3D simulator running on a Silicon Graphics IRIS workstation. 2) The development environment used for both the workstation and the on-board robot e.g. GNU software, MPW (for the Mac), X windows, MOTIF, VxWorks, CLIM, ... Operating system is standard Sun Unix for the perception and planning software, VxWorks for the real-time controller. User interface is done with a combination of X and Motif. Systems integration uses the home-grown Task Control Architecture (TCA), a message-based system for passing data and control information between distributed processes. Communication through TCA is via TCP/IP, and the architecture runs on Sun, NeXT, and RT machines. 3) Languages used: C, LISP, Prolog, CLOS, ... Development uses C exclusively. II. Mobile Manipulator 0) A brief description of your robot configuration. The Mobile Manipulator project is based on a HERO 2000, which is a 2-wheeled robot with manipulator arm and gripper. The sensors include scanning sonar, fixed base-mounted sonar, and a pointable wrist-mounted sonar. There is a BW camera mounted on the robot plus a ceiling-mounted camera for use within the laboratory. The tasks of the Mobile Manipulator include collecting cups from the lab floor, retrieving and delivering printer output, recharging itself when necessary, and exploring its surroundings. 1) Type of workstations (e.g. SUN, Silicon Graphics, IBM RS6000, HP, IBM PC/XT/AT/286/386/486, Macintosh, ...) used and why. If you more than one workstation, please say how they are networked (e.g. ethernet with TCP/IP, Appletalk,...) The on-board processors of the HERO are Z8088s (I think). These are programmed to execute guarded move commands and to provide a higher-level interface to the off-board system. The off-board system uses standard Sun workstations, mostly IPCs and ELCs. We also have a speech interface to the robot operating on a NeXT workstation. Communications to/from the robot is via a 9600 baud radio link. Camera images from the robot are transmitted via a standard Rabbit video transmitter, and digitized on a Matrox board. 2) The development environment used for both the workstation and the on-board robot e.g. GNU software, MPW (for the Mac), X windows, MOTIF, VxWorks, CLIM, ... Development is using standard Sun Unix. User interface is via X windows and text-based entry. Allegro CommonLisp is used for the LISP-based processes; the CC compiler for C-Based processes. Systems integration of the distributed processes is via the Task Control Architecture, which is used to connect both the LISP and C processes. 3) Languages used: C, LISP, Prolog, CLOS, ... On-board software development is in HERO-Basic (sigh). Off-board development is in C (for the controller and perception software) and LISP (for the planning and user-interface software). Date: Tue, 5 May 92 11:59:51 EDT From: slack@starbase.MITRE.ORG (M. G. Slack) Willie, I would like to do an informal survey of the development enviroments being used out there. The things that I would be interested in are: 0) A brief description of your robot configuration. A Denning MRV-1 mobile robot platform, -The base is synchro drive -An optional head platform is installed with an independent turn axis -upper ring of 24 sonars at 15degree separation about 2.5 feet off the floor -Lower ring of 6 sonars ~eq spaced with baffels about 5 inches off the floor -Pitch and roll clinometers -IR Beacon reading system -Laser target reading system -Base processor is 68000 which communicates to the robot's subsystems -Additional 100 Amp Hr 24 volt power and necessary Dc-DC converters to supply power to an on-board MacQuadra - We also have a PRISM real time sterio vision system which can be mounted on the robot and controlled to provide additional sensory information. 1) Type of workstations (e.g. SUN, Silicon Graphics, IBM RS6000, HP, IBM PC/XT/AT/286/386/486, Macintosh, ...) used and why. If you more than one workstation, please say how they are networked (e.g. ethernet with TCP/IP, Appletalk,...) The MacQuadra which is on-board uses RS232 ports to command and read the sensors and actuators leaving the lowlevel control to the periphial computers. Ethernet is used to down-load programs to the quadra which compiles them into a runtime module. There is also an RF modem which is used to send telemetry to another macintosh which is located remotely and allows a user to observe internal inforamtion of the robot as well as provides a mechanism for traded control. 2) The development environment used for both the workstation and the on-board robot e.g. GNU software, MPW (for the Mac), X windows, MOTIF, VxWorks, CLIM, ... Code is developed in a lisp environment (Mac Lisp or on a uExplorer) using the REX/GAPPS languages which reside inside of lisp. The result is a circuit definition which implements the developed algorithm. The circuit is then sent to a backend which generates a simulation of the circuit in C, Lisp, ... We use the C backend. The C code is then sent to the robot where it is compiled and run. 3) Languages used: C, LISP, Prolog, CLOS, ... C, C++, Lisp Date: Tue, 5 May 92 10:01:43 PDT From: gat@robotics.Jpl.Nasa.Gov (Erann Gat) Unfortunately, these are not short answers in our case. We have at least seven mobile robot systems and four manipulators and every one has a different setup - everything from Suns running RCCL to 6811's running ALFA. Perhaps if you told me why you need to know I could focus the answer appropriately. E. Date: Wed, 06 May 92 08:12:27 -0400 From: Reid Simmons [stuff deleted --- wlim] Reid PS In my description of the Ambler's controller, the two Motorola CPU chips are one 68030 and one 68020. Date: Wed, 6 May 92 10:22:17 EDT From: Kevin Dowling Willie, Most robot configurations here in Field Robotics and Vision and Autonomous Systems are mobile systems with some advanced manipulation work going on as well (Khosla et al). Nearly all workstations here are Sun SPARC's with a smattering of IRIS' for simulation and graphics although the Sun's running XGL are pretty fast. There are a couple of Macs too but are used only for some Mac applications and not for robot control etc. We do have special purpose hardware as well with a number of VME based systems, Maspar parallel processor, Titan graphics machine etc. Development evironments are UNIX usually running X11R5 and/or Openwindows environments. VXWorks is used for real-time development on most systems and CHIMERA (home brew RTOS) is used also for some manipulator control. There are extensively development systems for task control and communications such as the Task Control Architecture (TCA) and large libraries of imaging software (GIL) Language most used is C, not for any particular religious reasons but it is most convienent and offers HLL power and LLL flexibility. LISP is also used to a smaller extent. Hope this helps. nivek Date: Wed, 06 May 92 17:32:57 -0400 From: Chuck_Thorpe@IUS4.IUS.CS.CMU.EDU 0) Navlab and Navlab II mobile robots 1) Sun-4's, for historical reasons 2) X windows, sometimes Athena widget set 3) C ... but all the above may change in the near future, as we look for faster machines, look at C++, InterViews, etc. -- Chuck Date: Wed, 13 May 92 9:26:37 PDT From: Robert W. Albrecht eeb537 5-1600 Per your request, here is a summary of robot development environment. 0) ROBOTS: 2 Denning mobile robots with ultrasonics, laser reflectometer, video, tactile bumpers, infra-red beacon detection, wheel encoders, speech synthesis, etc. 1) WORKSTATIONS: 7 HP 9000 series 300 workstations - ethernet TCP/IP 2) WORKSTATION DEVELOPMENT ENVIRONMENT: Gensym - G2 real-time expert system in off-board workstations. radio link and/or wire line to robots. Locally developed multitasking operating system named LLAMA on board the robots running on MC 68000 in a basic OS-9 operating system. 3) LANGUAGES: Gensym - G2 off board with foreign functions in C. G2 is basically written in LISP. LLAMA on-board which is FORTH-like but written in C. -- Bob Albrecht From: maja@ai.mit.edu (Maja J. Mataric) Date: Thu, 21 May 92 10:48:11 EDT Robot configuration: We are working with 20 small mobile robots. The robots are rectangular foot-long wheeled bases equipped with piezo-electric bump sensors around the body, and a forklift for finding, carrying, and stacking pucks. The forklift has a suite of 6 infra-red sensors, two for obstacle avoiding, four in the jaw for puck positioning, and two for alignment and puck stacking. The system also includes two radio stations which allow the robots to triangulate their position, and transmit and receive 8-bit messages at 1Hz. This system allows us to test a variety of collective and cooperative behaviors. Workstations: Macintosh IIsi, Macintosh II. Development Environment: Macitosh Allegro Common Lisp. Languages: The robots are programmed in the Behavior Language, a high-level parallel robot programming language. The Behavior Language compiles into the Subsumption Architecture, which further compiles into a variety of target assemblers including HC6811 microprocessor assembler which is what our robots use. From: ian@ai.mit.edu (Ian D. Horswill) Date: Fri, 22 May 92 13:51:06 EDT Hi Willie, [stuff deleted --- wlim] I have an RWI base (B12) with a small VME card-cage on top. It contains a Pentek 4283 DSP card and a Data Translation frame grabber. It has a small Chinon CX-101 video camera with a 3mm lens and a voice synthesizer. A 6811 board acts as a peripheral processor. [stuff deleted --- wlim] Macintosh w/Appletalk and TCP/IP. Downloading is via the serial port. [stuff deleted --- wlim] My own lisp cross-compiler running under Macintosh Common Lisp v. 2.0. [stuff deleted --- wlim] Senselisp, a statically typed variant of Scheme with ML-style type inference, macros, and pointer arithmetic. The implementation does not support GC or closures for obvious reasons. -ian Date: Mon, 25 May 92 16:20:06 -0400 From: jar@cs.cornell.edu (Jonathan Rees) FYI: The Cornell Computer Science Robotics and Vision Lab has 2 mobots with: RWI 12" wheel base Gespak 68000 single-board computer (.5 Mby ROM, .25 Mbyte RAM) used for executive level control; multithreaded Scheme interpreter in ROM Custom Intel 80c196 board controlling various sensors IR proximity detectors and modem Polaroid sonar (12 transducers, with RWI control board) Speech, numeric keypad, LCD display, simple bumpers Scheme programs run autonomously, but are debugged with the help of Lucid on a workstation, attached with an RS232 tether (tether is used only for the debugging link). More mobots on the way: one with a custom tank-tread base, and one with vision. Jonathan Rees From: Liscano@IIT.NRC.CA Date: Tue, 26 May 1992 09:34 EST Organization: National Research Council of Canada. Project: EAVE Robot: Customized Cybermotion. Development: MAC II's, 68020's. Languages: C, HARMONY OS (multi-tasking & multi-processing), MacAPP. Ramiro Liscano Institute for Information Technology National Research Council Bldg M-50 Montreal Rd. Ottawa, ONT. Canada K1A 0R6 Tel: (613) 993-6565 liscano@iit.nrc.ca Date: Tue, 26 May 92 10:53:15 -0400 From: jar@cs.cornell.edu (Jonathan Rees) And for what it's worth, there's a paper describing the Cornell mobot and its development environment: Jonathan Rees and Bruce Donald. Program mobile robots in Scheme. {\em Proceedings of the 1992 IEEE International Conference on Robotics and Automation,} pages 2681-2688. Date: Mon, 08 Jun 92 23:01:18 -0500 From: Avi Kak Willie, Looks like you completely missed out on Purdue University in your comp.robotics survey of May 23, 1992. The mobile robot PETER in the Robot Vision Lab at Purdue University is one of the earliest K2A robots made by what was then known as the Cybermation company. This robot is capable of completely autonomous indoor navigation using model-based vision. The robot is capable of maintaining a speed of approximately 10 to 12 meters per minute under vision control. This speed is not affected much by collision avoidance functions. Collision avoidance is done by using ultrasonic sensors. Further details on how the robot does this can be found in the forthcoming article Authors: A. Kosaka and A. C. Kak Title: Fast Vision-Guided Mobile Robot Navigation using Model-based Reasoning and Prediction of Uncertainties Journal: Computer Vision, Graphics, and Image Processing -- IMAGE UNDERSTANDING, September 1992 (To appear). Attached below are the vital statistics. Avi Kak kak@ecn.purdue.edu Organization Project Robot Development Languages & Type HW Environment SW Environment ============= ============= ==== ============== ============= Purdue Navigation Cybermotion VME 68030 VxWorks, C University and K2A Control using Vision Date: Tue, 09 Jun 92 11:52:23 -0500 From: Avi Kak Willie, We use a Sun4 for software development. All the software is in C. Up to this time, the camera images were transmitted to the SUN4 over a video wireless link and all the vision processing was done off-board. The collision avoidance was done on-board by a 68030 supervisory processor. The supervisory processor ran a XINU microprocessor operating system developed by Prof. Doug Comer of the CS Department here. Last February we purchased a VxWorks operating system for the 68030 on-board computer. Our plan this summer is to do all processing on board and that includes all the vision processing. You might also like to know that the mobile robot has a personnel detection system that uses a combination of thermal and ultrasonic signals in an on-board neural-network based procedure. For up to 8 feet, this system has no false alarms and 100% detection accuracy. We are currently trying to increase the range of robust personnel detection, since 8 feet is simply not enough for most sponsors. Incidently, our mobile robot project is now seven years old. Being not too PR-oriented, we simply have not made too big a noise about it. We did present our work at this year's AAAI Spring Symposium. Avi P.S.: PETER stated drooling at the prospect of meeting SmartyCat. Perhaps a meeting could be arranged in the not too distant future. Date: Tue, 30 Jun 1992 13:18:02 +0300 From: Virpi Santti You asked for information about research going on among robotics about a month ago. I received this request only yesterday, but I hope You still are interested.. Well, here it comes: Akseli is the official nickname given to the autonomous mobile robot constructed in the Machine Automation Laboratory, Technical Research Center of Finland (VTT). Akseli has a self-made base with a turtle configuration: On each side there is an indepen- dently driven wheel, in the back and front there are free castor wheels. The drive wheels are controlled with HP-1100 microchips. Akseli carries now a 386-PC running MS-DOS. In the near future the operating system will be substituted by LynxOS. For the vi- sion system Akseli carries 16 solid and 5 rotating ultrasonic sensors, and 6 infrared sensors. Akseli has a radio modem link to a controlling (and monitoring) computer. The software is generat- ed using C-programming language. Akseli is built for testing different kinds of algorithms for path planning, navigation, movement control, environment percep- tion and collision avoidance. If You do need more information, do not hesitate to contact. Regards, vjs. ****************************************************************************** Virpi J. S"antti Internet: Virpi.Santti@kau.vtt.fi Research Scientist or, vjs@kau.vtt.fi Technical Research Centre of Finland Tel. +358 31 163 630 P.O. Box 192, SF-33101 Tampere, Finland Telefax +358 31 163 494 ****************************************************************************** Date: Sat, 18 Jul 1992 17:51:23 -0400 From: rg@nymph.msel.unh.edu (Roger Gonzalez) Here is another belated addition. The MSEL robots are underwater autonomous vehicles (AUVs). Two are open-frame high precision short range testbeds, one is a hydrodynamic long range long endurance data sampling vehicle. Our old robots run low-level code using pSOS on 68000 boards built at the lab. High level applications run on Ironics 6U VME boards also running pSOS. The two levels communicate via a serial line. Code is developed for the low level system on a Charles River Data System Universe 68, and for the high level on an Ironics Performer 32. Code is burned into prom for the low level, and stored on a WORM drive for the high level. In the lab, code can be dynamically downloaded to the low level via serial lines, and to the high level via a Bit 3 VME-VME adaptor. This setup is pretty old, and thus we are moving to.... Two VME card cages running low power Oettle+Reichler CMOS boards, connected via ethernet. One cage is for vehicle system software, and is 3U format. The other cage (optional) is for high level applications, and is 6U. Both run VxWorks, and communicate with their Sun Sparcstation development environments via ethernet. -Roger Date: Sun, 19 Jul 92 15:57:25 -0400 From: connolly%rabbit@cs.umass.edu (Christopher Ian Connolly) [stuff deleted --- wlim] Well, I thought I'd add to the list and give you a brief description of the mobile robot work here at UMass: We (the Laboratory for Perceptual Robotics) have a Denning platform with a 24-transducer sonar ring - I don't know the model number offhand. It's due to be gutted anyway to put in a VME cage with our own software, so I would imagine a model number isn't too important. We use it as a testbed for reactive planning using harmonic functions, which in turn serves as our model for motor planning in the mammalian central nervous system. It will eventually be equipped with a stereo head, but for the moment only has sonar. A DECstation 5000 is normally used for control and planning. The computer vision group here also has a Denning equipped with sonar ring and camera, which they use for (among other things) visual navigation experiments. In conjunction with the vision group, we are also experimenting with real-time path planning (again using harmonic functions) where the obstacles are visually obtained. In this case, a Sun sparcstation is the host. In both cases, we prototype in Lisp and C, and write the end code in C. -CC -- - - - - - - - Christopher Ian Connolly connolly@cs.umass.edu Laboratory for Perceptual Robotics wa2ifi University of Massachusetts at Amherst Amherst, MA 01003 From: Gregory Dudek Date: Wed, 22 Jul 1992 10:27:34 EDT Hello, I hope you're well. The mobile robot survey was good to see. I thought I might send you info on our lab here at McGill University's Research Centre for Intelligent Machines (McRCIM). 0) ROBOTS: RWI B-12 robot with 13 sonar sensors; other robots and sensor systems are under development. We should have a pan-tilt video head working with it soon. There's also work being done on walking robots. (We also have a variety of non-mobile robot arms.) 1) WORKSTATIONS: Mostly SUN Sparc machines. Networked to other machines (SGIs, MASPAR, Datacube) via ethernet & TCP/IP. Some mc68hc11 devices for low-level control functions. 2) The main development environment is based on Standard SUN UNIX tools and some GNU stuff, all using X-windows. Small-C on IBM PC is used for the 68hc11s. 3) Pretty much all our work in done on C and C++. Regards, Greg Dudek. Date: Thu, 23 Jul 92 14:45:14 BST From: Ulrich Nehmzow Willie, I picked up your review of current robotics programmes today, and I wonder whether you would be interested to include the Edinburgh robots. At the Department of Artificial Intelligence at Edinburgh we have built two small mobile robots to conduct experiments in the autonomous acquisition of motor-sensory and navigational skills: the robots learn through experience and use connectionist computing architectures to associate sensor signals with motor actions. There are a number of publications, for example in "From Animals to Animats", MIT Press 1991 and the proceedings of "Intelligent Autonomous Systems 2", Amsterdam 1989. ALDER Alder is the first robot we built, it's base is made from Fischertechnik (a technical kit), it has two driving motors and a caster wheel. The controller is an ARC52, using an 8052 microprocessor. The robot has up to eight binary sensors (tactile, revolution counter, forward motion sensor and pushbutton switches) and one ultrasonic sensor. It is programmed in BASIC, programs are downloaded from either a SUN or a PC. CAIRNGORM Cairngorm's base is also made from Fischertechnik. The robot has a Flight 68k controller, using a 68000 processor. Programs are written in C on a SUN, crosscompiled and then downloaded to the robot. Cairngorm has binary sensors (tactile, revolution counter and pushbutton switches). The work will now continue at the Laboratory for Cognitive Neuroscience at the Department of Psychology, here in Edinburgh. I hope this is usful for you, please ask if you want more information! Ulrich. Ulrich Nehmzow Laboratory for Cognitive Neuroscience at the Department of Psychology Edinburgh University Scotland ulrich@castle.ed.ac.uk Date: Fri, 1 Jan 93 11:37:01 EST From: murphy@baby_doe.mines.colorado.edu (DR. ROBIN R. MURPHY) I'd like to add our robotics program to the list: Organization: Colorado School of Mines Project: Mobile robots Robot type: Denning MRV-III with Sparc IPX directly mounted Development/HW Environment: Sparc IIs, IPXs, IBM RS/6000s Languages and SW Enviroment: C, X11, Khoros, potential fields X11 visualization tool (homemade) Other information: Our basic research approach is a hybrid hierarchical/reactive architecture, similar to the AuRA architecture. Our mission is: The Mobile Robotics/Machine Perception Laboratory is a facility devoted to basic and interdisciplinary research, technology transfer, and hands-on education in artificial intelligence through robotics. Research and technology transfer efforts will concentrate the reduction of human risk in hazardous situations, stewardship of the environment, and/or improvement of the quality of life through better manufacturing processes. Our specific projects are on robotic navigation for situation assessment and rescue operations (think "mine cave-in" and Chernobyl), transportation of hazardous materials, and space mining. Colorado School of Mines is a special purpose university for science, engineering, and high technology, located in the foothills of the Rocky Mountains 20 minutes from Denver. We have approximately 3000 undergraduates and 1300 graduate students and 230 faculty. CSM offers BS, MS, and PhD degrees in Mathematical and Computer Sciences, as well as an interdisciplinary Robotics and AI minor. Contact person: Robin R. Murphy, Assistant Professor Department of Mathematical and Computer Sciences rmurphy@mines.colorado.edu Thanks, Robin Date: Fri, 16 Jul 93 14:49:20 PDT From: Kurt Konolige FLAKEY Artificial Intelligence Center, SRI International Flakey is an autonomous mobile robot used for research in control, planning, sensing, and acting. Flakey's control software is built on modules called control structures, which incorporate a set of fuzzy rules designed to reliably achieve a particular goal in a specified context. Multiple control structures operate in parallel, and are coordinated by a task-level planner. Flakey integrates several diverse methods of acquiring and interpreting sensor data, including an occupancy grid for local obstacle avoidance based on sonars and stereo vision; 2D recognition routines for landmark acquisition and tracking; and geometrical modeling procedures for recognizing and tracking objects. Flakey is a homebrew base incorporating 2 golf-cart gel cells for a total of 700 watt-hours of energy. Typically we can run for 3 hours without recharging, more if there is relatively little motion. Flakey is equipped with sonar and bumper sensors and two video cameras. Computation is supplied by a Z80 motor controller and a Sparc10/30 processer, upgradeable to 4 processors. The Z80 provides basic motor control, odometry, and sonar firing; cycle time is 20ms. The Sparc handles everything else, including speech synthesis and recognition, vision and sonar interpretation, planning and reactive control. There is a radio ethernet bridge (AGILIS) at 200kb to our Sun network for development. The Sparc has a standard UNIX environment (no special realtime software) running Lucid LISP and C. Most of our software is written in Lisp, including the realtime controller that operates at 10Hz. Graphic output is with X-windows. There is a simulator which is heavily used for debugging. From: Tyson David Sawyer Date: Tue, 20 Jul 1993 17:16:36 -0400 0) a) James b) Worcester Polytechnic Inst. c) Develop methods for autonomous robotics d) Just beginning e) Vision and navigation 1) James is built on an RWI B12 mobile base which has an 8 bit 6MHz NEC 78310 uController. It is connected by an RS-232 to an RWI sonar controller which has a 68HC11. The 68HC11 acts as the central behaviour controller in its spare time after controlling the sonar. There are 12 sonar transducers evenly spread over the front 180 degrees. The B12 base is equipped with contact sensors. All programming was done in assembly and Small-C. 2) A Gateway 2000 type PC was used for all programming. 3) Development was done under windows, but all the compilers and assemblers ran under DOS. 4) Languages used: 78310 assembly, 68HC11 assembly and small-C From: masaki@ai.mit.edu (Masaki Yamamoto) Date: Tue, 20 Jul 93 18:57:55 EDT Hi Willie, There are two robots that I would like to make entries. --Masaki Yamamoto (Visiting scientist from Panasonic at MIT AI lab) ********************************************************************** SUGGESTED FORMAT FOR PROJECT DESCRIPTIONS: 0) A brief description of your MOBILE robot project: a) Project/robot name SOZZY b) Institution MIT c) Goals of the research To show a new prototype of a domestic robot: 1) Small and cheap robots which can be in the middle of conventional service robots and hobby robots. 2) Robust performance based on insect-like behaviors. 3) Simulated hormone system which makes the robot behavior more flexible and sometimes more human-friendly. d) Current state Programing and experimenting. e) Future plans Finishing the project by writing a paper on it. 1) Robot configuration: type of sensors: IR, bump switches, dust sensor, pyro sensor on-board processors: 6811 mobile base: "homemade" 2) Type of workstations: Macintosh as Behavior Language requires it. 3) The development environment used: Macintosh Common Lisp 4) Languages used: Behavior Language ********************************************************************** SUGGESTED FORMAT FOR PROJECT DESCRIPTIONS: 0) A brief description of your MOBILE robot project: a) Project/robot name GOPHER b) Institution MIT c) Goals of the research To show a new prototype of a office robot: 1) Compact and cheap robot which can navigate through narrow space and can be safe in case it should fail. 2) Vision based navigation using newly developed a compact, low-power, real time, active vision system. d) Current state Programming and experimenting. e) Future plans Preparing a paper. 1) Robot configuration: type of sensors: active vision sensor with 2 DOF, IR, tactile sensor, flux gate compass on-board processors: 68332 bard (Vesta) mobile base: R2 (ISRobotics) 2) Type of workstations: Macintosh, SUN as Behavior Language and GCC requires them. 3) The development environment used: Macintosh Common Lisp, X windows 4) Languages used: Behavior Language, GCC From: Gregory Dudek Date: Wed, 21 Jul 1993 14:37:58 -0400 SUGGESTED FORMAT FOR PROJECT DESCRIPTIONS: 0) A brief description of your MOBILE robot project: There are several projects. These phrases subsume several: 1 - autonomous exploration, sensing and map construction at multiple levels of abstraction. 2 - multi-sensor active vision for mobile robotics. a) Project/robot name 1) The Mobile Robotics Project (contact: Dr. G. Dudek). 2) The QUADRIS project (contact: Dr. M. D. Levine). b) Institution McGill Research Centre for Intelligent Machines McGill University c) Goals of the research To develop systems capable of learning about and acting in an a priori unknown environment. In the short term, to develop models for the respresentation of space, for intelligent exploration, and for scene understanding using multiple sensors. d) Current state Vision and robotics groups (including several faculty each) are pursuing several topics independently on several fronts (biological vision, 3-D modelling, manipulator design, etc.) The mobile robotics group is pursuing a combination of theoretical issues and practical problems. The practical testbed is a small commercial mobile platform. e) Future plans The scope of our research is currently expanding, both in terms of research goals and hardware. A new lab has almost been completed. Ongoing long-term research objectives are too varied to be fairly summarized here. 1) Robot configuration: type of sensors (e.g. sonars, feelers, "vision", IR, ...), on-board processors (e.g., VME boards, Z80's, 486's, 6.270 boards, mini-boards, ...), mobile base (e.g., "homemade", Cybermotion, RWI, TRC, Denning, Nomadics,...) Base: RWI B-12. Sensing: Combination of sonar, video (mounted on a pan-tilt head), and feelers. An active head and active laser sensing hardware is under development. On-board processing: minimal; a couple of microcontrollers and, sometimes, a 68000-based system. Cost computing is performed off-board. 2) Type of workstations (e.g., SUN, Silicon Graphics, IBM RS6000, HP, IBM PC/XT/AT/286/386/486, Macintosh, ...) used and why. If you have more than one workstation, please say how they are networked (e.g. ethernet with TCP/IP, Appletalk,...) SUNs are used as general purpose comute engines. SGIs are used for "virtual reality" controllers (and general computation). A Macintosh is used to provide portability and presentation design. Everything is networked via ether using TCP/IP. Some vision related stuff is being developed for C-40 microcontrollers. 3) The development environment used for both the workstation and the on-board robot e.g. GNU software, MPW (for the Mac), X windows, MOTIF, VxWorks, CLIM, ... The usual stuff for the workstations: X-windows, IRIS GL. 4) Languages used: IC, LISP, Prolog, CLOS, ... Mainly C. Return-Path: Date: Mon, 26 Jul 1993 19:53:06 +0900 Our mobile robot project Hiroshi Ishiguro & Saburo Tsuji Department of Systems Engineering, Osaka University Toyonaka, Osaka 560, Japan 0) A brief description of our mobile robot project: a) Robot name: SAI (Situated Artificial Intelligence) b) Institution: 1992 .... c) Goals of the research: (Goal 1) Development of an autonomous mobile robot which can recognize an unknown environment with panoramic sensing. (Goal 2) Development of a vision-guided mobile robot which can accommodate to a dynamically changing world with multiple vision agents (4 vision sensors). d) Current state: (Goal 1) The robot can move at a real time in an unknown environment only with vision sensors and build environmental models. (Goal 2) The robot can simultaneously observe several events in a dynamic world. e) Future plans (Goal 1) Making qualitative maps of the environment (Goal 2) Development of an attention control system for the multiple vision agents. 1) Robot configuration a) Type of sensors: 4 vision sensors which can be rotated independently around a common axis. b) On-board processors: VME board computer (CPU=68030) c) Mobile base: "Homemade" (two driving wheels with optical encoders) 2) Type of workstations a) Workstation: SUN workstations (IPX, Spark station 2) b) Computer network: Ethernet with TCP/IP 3) Development environment a) Workstation: X windows b) On-board computer: VME computer debugging monitor 4) Languages used: C Date: Mon, 19 Jul 93 10:20:44 EDT From: bartha@corsair.aa.wpafb.af.mil (Bartha) 0) MOBILE robot project description: a) Project/robot name: no name yet b) Institution: Wright Laboratory c) Goals of the research: Learning to intercept targets using drive-reinforcement neural networks. d) Current state: Enough hardware in place to begin development and testing. Development environment still being set up. e) Future plans: Incrementally add new behaviors. 1) Robot configuration: HERO 2000 robot with rotating and base sonar, light, sound, temperature, and battery sensors. A CCD camera for vision and additional ADCs for tactile sensors will also be added. 2) Type of workstation: IBM compatible 286 which will soon be replaced by an onboard 486 notebook computer. The main reason an IBM PC based system is being used is that the existing vision processing hardware and software is IBM PC and DOS based. Low cost is another reason. 3) The development environment: DOS and HERO BASIC 4) Languages used: Microsoft C and Assembler Dr. Gabor Bartha Wright Laboratory (AAAT-1) Bldg 635 2185 Avionics Circle WPAFB, OH 45433-7301 FAX: 513 476 4302 Phone: 513 255 7647 Date: Mon, 16 Aug 1993 15:54:51 -0400 (EDT) From: jdc@meceng.coe.neu.edu (Jill D. Crisman) Willie, I'm sorry for the late response to your request for "brief" descriptions of mobile robot projects. I hope that we are in time to be added to the compliation of the list. We currently have two mobile robot projects: _______________________________________________________ The first project: a) Our robot is called Phaeton b) We are the Robotic and Vision Systems Laboratory (RVSL) at Northeastern University. c) Our long term objective is to develop autonomous, general-purpose robotic systems by first developing general-purpose semi-autonomous systems, then incrementally decreasing the system's dependence on human interaction. Our current focus is on an interactive, general-purpose, mobile robot control architecture which is natural and intuitive for humans to use. Immediate goals include the discovery of necessary and sufficient command primitives for interactive navigation by deixis (pointing). d) We have developed an simple control architecture consisting of six control loops and have shown that this architecture is stable. We are currently working on: a simulated system to investigate a necessary and complete set of deictic primitives. implementation of deictic primitives on a robotic system development of a future general-purpose wheelchair robot test platform. e) The visual tracking and basic behaviors will be combined to allow the robot to follow targets. Then deictic primitives will be designed/discovered to allow interactive control of the robot via its digital camera images. Ultimately a wheelchair with on-board power will provide a comprehensive platform allowing a user to take the robot for a spin without much effort. The robot's sensory capacity will be enhanced over time. 1) Robot configuration: Robot configuration: We have a Denning MRV3 base with a ring of 24 Polaroid ultrasonic sensors and an on-board 68000 which controls the sonar and motors. Two pairs of CCD cameras are used at various times: monochrome (Cognex) and color (Pulnix). Two independent camera controllers (Directed Perception) provide pan and tilt. A rotating platform will be built to track the wheels' direction since the Denning's synchronous steering results in the body never turning. Vision processing is provided by a monochrome Cognex 4400 and color Datacube boards (Digicolor and Framestore); both are VME-based. 2) Type of workstations: The primary processor is an off-board Sun 4/330 connected by serial-line tether to the Denning. Some simulations are done on Macintosh computers. 3) The development environment: Development environment: On the Sun, X and emacs provide a basic environment. The Cognex is programmed via a PC-based front-end using epsilon. 4) Languages used: C on the Suns, Macintosh and Cognex. _______________________________________________________ The second project: a) Our robot is called the Lobster Robot b) We are the Robotic and Vision Systems Laboratory (RVSL) at Northeastern University. c) Our long term objective is to develop a shallow water walking robot based on biological models of lobsters. This is particularly challenging due to the surge and surf action of shallow water. d) We have developed a biologically based control architecture and have shown in simulation that this architecture produces neural patterns observed in walking lobsters. We are currently working on: a simulation to investigate kinematics and dynamics of an eight-legged underwater vehicle, implementation of the central pattern generator on a single legged robot, and development of our future 8-legged underwater walking machine e) Our future work includes the implementation of our biologically based control architecture on an eight-legged underwater platform. We are also studying lobster behaviors which allow them to ambulate in hydrodynamic environments and integrating these behaviors into our current control architecture. 1) Robot configuration: We are currently designing our eight-legged robot system. At present, we are using a single legged system which we have designed and constructed. The leg has 3 DOF, each driven by DC motors under PWM control. The leg has an HC11 based single-board computer that controls its joints. 2) Type of workstations: Simulations are done using Macintosh computers. 3) The development environment: We are currently programming our single-legged robot in assembler using the environment provided with the single-board controller. We are currently investigating which development environment we will use for our eventual system. 4) Languages used: C, Pascal, and Assembly. Date: Fri, 29 Oct 1993 09:06:07 -0400 From: Frank V Koss I hope it's not too late, but I have a submission from the University of Michigan. If you can't integrate it with the others, my loss. Frank Koss Artificial Intelligence Laboratory University of Michigan ***** 0) There are a number of projects involving mobile robotics at the University of Michigan. ***** I) a) CARMEL (Computer-Aided Robotics for Maintainance, Emergency, and Life-support) b) University of Michigan AI Laboratory and Mobile Robot Laboratory c) Basic research into the use of mobile robots in nuclear reactors (funded by the DOE): navigation using cognitive maps, sensory-based plan recognition, planning, high-speed obstacle avoidance d) Obstacle avoidance research has concluded successfully, planning research is on-going. Coordination research is on-going. Navigation research has reached quiescence with the defense of a thesis. e) Upgrade on-board processing to allow for multiprocessing, cooperation with BORIS II) a) BORIS (Box Organizing Robotic Intelligent System, due to its task in the AAAI 93 Robot Competition) b) University of Michigan AI Laboratory and Mobile Robot Laboratory c) Basic research into the use of mobile robots in nuclear reactors (funded by the DOE): high-speed obstacle avoidance, research into dead reckoning, coordination of multiple robots, planning, navigation d) Close to a usable research platform, hardware is stable, software is coming along. Obstacle avoidance research concluded successfully. e) Upgrade on-board processing, cooperation with CARMEL III) a) MAVERIC (Michigan Autonomous Vehicle for Experimental Research in Coordinated exploration) b) University of Michigan AI Laboratory c) Basic research into coordination of agents in the field (funded by ARPA): outdoor navigation, mantainance of large evironment models and the assimilation of information, coordination of multiple robots d) All hardware and software is in place for autonomous navigation using vision. Navigation incorporates road/trail following and obstacle avoidance. e) Making improvements in the low level controller, and have begun the construction of the second vehicle (MAVERIC 1.0.1). Will add more sensors (sonars, laser range camera, stereo grayscale cameras). ***** 1) ***** I) CARMEL Sensors: ring of 24 ultrasonic sensors, grayscale video camera On-board processors: 1 80286-based PC for firing and processing ultrasonic sensors, 1 80486-based PC for higher-level functions (obstacle avoidance, planning, vision, etc.) Base: Cybermotion K2A II) BORIS Sensors: half-ring of 12 ultrasonic sensors, grayscale video camera On-board processors: 80486-based PC Base: TRC Labmate III) MAVERIC Sensors: color video camera, positional encoders Processors: 1 Sun Sparc 10, 1 80486-based PC, Datacube MaxVideo 20 Base: "homemade" from an electric utility vehicle. Six 6 volt batteries drive a 36V DC motor. We have on board 120V AC from an inverter. ***** 2) ***** The projects at the University of Michigan have thus far adhered to a principal of having all processing on the robot. Difficulties with communication links are avoided. Also, the complete development environment is always available, even when the robot is "in the field". However, various lab workstations (DECstations, Silicon Graphics, Sun Sparcs, IBM RS/6000s) are used for some very preliminary development of algorithms. I & II) CARMEL and BORIS PC-compatible workstations (running MS-DOS) are used due to their expandibility, simplicity, and low-cost. High-speed obstacle avoidance requires that the processing be very close (temporally) to the sensors and actuators. An on-board PC avoids communication delays over wireless connections, and DOS supports only a single process at a time which cannot be pre-empted by another process, network activity, etc. CARMEL's three computers communicate over RS-232 serial lines. The 80486 computer requests sensor readings from the 80286 and sends commands to the base. BORIS's PC also sends commands to the base over an RS-232 serial line. III) MAVERIC The larger size and battery capacity of MAVERIC allows for more on-board processing power. The PC-compatible workstation is used for low-level vehicle control, while the Sun workstation and MaxVideo 20 carry out planning, sensor processing, etc. The PC, Sun and MaxVideo 20 (VME chassis) receive power from the AC inverter. The Sun and the PC communicate over an RS-232 serial line. The Sun communicates with the MaxVideo 20, which is housed in a VME cage, over an Sbus-to-VME translator. ***** 3 and 4) ***** I & II) CARMEL and BORIS All of the software for these two robots is written in C using the Borland C++ 3.1 compiler, except for the ultrasonic sensor firing and processing software on CARMEL which is written in FORTH (and has not changed in over two years). All PCs run MS-DOS. No simulators of any sort are used. III) MAVERIC We run SunOS4.1.3 and the MIT X-Windows (with twm or mwm). Software on the workstation is all written in C using gcc 2.4.5. The MaxVideo 20 is controlled from the Sun workstation and is programmed using Imageflow 2.2.6. The 80486 PC runs MS-DOS with all software written in C using Borland C++ 3.1. From: bridget@aifh.edinburgh.ac.uk Date: Fri, 29 Oct 93 13:50:10 GMT Here at Un of Edinburgh we also have (apart from the FischerTeknik robots you already mention) two mobile RWI-base robots with transputer facilities which run autonomously with programs downloaded from PCs (`Bill' has a PC motherboard on-board, `Ben Hope' just uses a transputer as its controller). These are programmed in parallel C or ANSII-C with transputer/parallel extras. Then there are the student vehicles, made of LEGO with 68000 processors on. Programmed in C or our custom-built LISP-like language CPL (Control Process Language). Then there are the robot arms -- an ADEPT, several RTXs, a parallax robot (which we've not got very far with), an armdroid and a few other bits and pieces. Students program in another custom language from PCs; researchers program the ADEPT in VAL2 and Prolog; RTXs in turboPascal normally, from SPARCs. I hope that this is all the info you need to update our entry. I would hate everyone out there to think that two FischerTecknik vehicles is all we work with! ----------------------------------------------------------------- Bridget Hallam Dept of Artificial Intelligence, _______ 5 Forrest Hill, | |___|___| Edinburgh EH1 2QL |-|_______|-| Scotland O O Children know how love is spelt --- t - i - m - e. Date: Fri, 29 Oct 1993 15:59:44 -0400 From: Patrick Kenny Hey you didn't get us in your survey. University of Michigan, AI and Robotics Lab Robot Type: Indor - CyberMotion K2A, TRC Labmate, 6811 Minirobots Outdoor - Golfcart, 4 wheel base. HW Development: SUN Sparc II, 10; PC 386, 486, 6811 SW Development: X, c, c++, 6811, Lisp, InterViews, DataCube, Gnu I forgot, we also have an Andros base, A treaded tank like vehicle. Questions to: pkenny@eecs.umich.edu Thanks, -pk From: pwelliso@major.cs.mtu.edu (PETER W. ELLISON) "Hello my name is Tracy" _____________ | | |___| | | /------------------------\ | ______________________ / |O O/ \O__O___O_____O___O__O/ "I'm from Michigan Tech. The IEEE local chapter is my caretaker I used to have a 6502 onboard computer, capable of small amounts of autonomous movement, for example go farward for 1 second. The software from the looks of it was developed on a Apple IIe. There is a remote control station that has a Apple IIe that interperates the joysticks and sends commands over the 1200 baud link. This system came with no docs and the software had been deleted so ... IEEE developed an onboard 68hc11ebv system that talks via direct connect to a rs232 to a IBM PC. the 68hc11 software was developed on a SUN UNIX work station cross assembled and put on the 68hc11. This system is capable in theory (we ARE going to enter the Unguided Robot Comp) of total unguided operation. The onboard OS is custom (it does VERY little) for the most part it is a large serial to parallel converter. The PC will be running MS DOS 6.0 :( I wish it was otherwise but that is life. The OS will play a very small part in the over all life of me." Short Version: location: org. onboard OS onboard cpu MTU.EDU IEEE Custom OS 68hc11ebv in addition there is a onboard PC that talks to the 68hc11 running dos6.0 all software is created and worked on SUN SPARC workstations the ported to where ever it is nessary. (I'm a CS major that likes UNIX :) Peter Ellison -- / pwelliso@major.cs.mtu.edu *####*---------------------------------------------- \ All flames etc. to /dev/null. --=}:) Date: Sat, 30 Oct 1993 14:55:25 -0400 From: Patrick Kenny Looks fine, you can add at the top, that we have several 6811 based minirobots, and an Andros (not nammed yet) base robot. Thanks, -pk P.S. I didnt have the latest version, sorry. Date: Thu, 9 Jun 1994 12:58:26 -0400 From: Jonathan Monsarrat Hi! Ok... not having seen the survey, I'm not exactly sure what you want. Brown has 4 robots right now... Gort has a big rwi base, onboard 486 with sonars, IRs, and laser striper Ramona has a big rwi base, onboard 486, sonars, IRs, camera Arms are being built for both our big robots. We have two small robots based on small RWI bases. We do our programming in either C++ or Lisp and Rex. Could you send me the current version of the survey? Then I can give you information fitting the "style" of what you have. Thanks -Jon %! Jon Monsarrat jgm@cs.brown.edu moderator, comp.sources.postscript %! 9 9 scale 9 9 moveto(qll-??LHHL??llH?hH7t,7olCAHH@){dup 10 mul rotate 80 lt{4 0 rlineto}{4 0 rmoveto}ifelse}forall stroke showpage From gedney@charlie.ece.scarolina.edu Tue Nov 29 10:47:45 1994 Return-Path: From: gedney@charlie.ece.scarolina.edu (Charles A. Gedney) Subject: Re: Robots To: wlim@lehman.com (William Lim) Date: Tue, 29 Nov 94 10:47:00 EST In-Reply-To: <9411281838.AA00570@fermat.lehman.com>; from "William Lim" at Nov 28, 94 1:38 pm X-Mailer: ELM [version 2.3 PL11] Project Description for Cybermotion K2A Large quantities of radioactive Low-Level Waste (LLW) and Transuranic Waste (TRU) are currently in storage at a number of sites throughout the United States. Environmental Protection Agency (EPA) regulations call for regular inspection of such storage areas requiring significant increases in manpower. The overall objective of this work is to develop a semi-autonomous vehicle capable of inspecting a low radiation level contaminated storage area of stacked drums or other containers aligned in aisles. The vehicle will have the capability of autonomously entering, retracing the entry route, and avoiding all hazardous obstacles in the route traveled. Phase 1 In Phase 1 the currently available K2A served as a test bed for the development of improved navigation and machine vision algorithms. Using current technology, machine vision algorithms will be developed to identify storage drums and detect variations in the condition of the storage containers. Phase 2 During Phase 2 a new mobile platform, the K3A, will be developed and a light- weight arm will be placed on the K3A superstructure to meet several surveying requirements. Automated machine vision will be used during routine patrol of storage areas to detect potential problems with storage containers. Acquired data will be maintained in a database suitable for meeting EPA reporting requirements. After suitable demonstrations of all robotic capabilities in Phase 2, Phase 3 will focus on the manufacture of these enhanced intelligent robots. Project being funded by the Department of Energy (DOE) and work is being done at the University of South Carolina and Clemson University. Cybermotion NAVMASTER K2A * On/Off-board Host Computers * Ultarsonic Imaging System * Docking Beacon System * Battery Charger * Modem Link * Joystick * Teach Pendant Mechanical * 3-wheel, synchro-drive * DC steering Motor with 106:1 spiral gear reducer & optical pulse encoder * DC Drive Motor with 24:1 gear reduction box & optical pulse encoder * Turret that turns in the direction of the wheels * Slip Ring that provides electrical connections between base and turret * Front bumper has kill switch Electrical * Control Computer, Z-80 CMOS CPU * Power amplifiers to drive the motors, optically isolated signals from computer * LED's for system status * Power system: 2 - 12v. (65 AHr) on-board batteries in series, power converters provide +5, -12, and +12 for electronic circuits. * RS-232 communications between computers * Turret Interface Panel (TIP) provides routing of RS-232 lines * Piezoelectric transducer - Ultrasonic Imaging System * Manual Teach Pendant Programming * Assembly Language (PASM) Self Guided Vehicle Control Language 64 Instructions * Robot Language ------------------------------------------------------------------------------ Project Description for Heathkit Hero 1 ET-18 We aren't actually doing a project with Hero, he is being used in our Robotics class as a model so that the students can write kinematic equations. A compiler was written to generate a S-record that can be downloaded to Hero. Mechanical * Arm, wrist, turret Electrical * Motorola 68000 microprocessor Programming * Assembly from the keypad * GEHPL - Gedney England Hero Programming Language Written for the DOS/Windows and UNIX OS's