Newsgroups: comp.robotics
Path: brunix!news.Brown.EDU!qt.cs.utexas.edu!yale.edu!newsserver.jvnc.net!gmd.de!Germany.EU.net!mcsun!sun4nl!fwi.uva.nl!smagt
From: smagt@fwi.uva.nl (Patrick van der Smagt)
Subject: Re: 3-D Xwindows robotics program
Message-ID: <1993Feb9.102853.4057@fwi.uva.nl>
Sender: news@fwi.uva.nl
Nntp-Posting-Host: brad.fwi.uva.nl
Organization: FWI, University of Amsterdam
References: <1993Feb8.160957.1@cajun>
Date: Tue, 9 Feb 1993 10:28:53 GMT
Lines: 37

daship@cajun writes:

>I'm thinking about writing a program that emulates robots of different
>configurations, and their mechanics.  You would be able to enter the
>number of joints, what type of joint each is, the length of connecting
>arms between joints, the initial angle of joints, etc.  I would probably
>implement a parameter table to do this.  Anyway, after entering this
>information, you could specify desired final angles and the program would
>show the movement of the robot in three dimensional graphics (hopefully in
>real-time wireframe).  I would like to write the program in C++ and X-windows.
>My question is this... are there currently any programs out there that I
>would be duplicating the effort of?  I would also be interested in knowing
>what you'all think would be a good platform for the development of this
>program (to be as portable as possible... I have access to Sun, HP, PC, and
>possibly DEC workstations).  I would appreciate any comments.

Yes I wrote such a program for X windows.  In fact, it consists of
three programs: one controller (which YOU should write), one kinematics
simulator (calculates position, speed, and acceleration of the joints,
both in joint and Cartesian domain), and a 3D display program.  The
whole thing is rather modular, and documented; yet I always have the
feeling it's not 100% finished.  The graphics is poor (line drawings),
since I wanted it to be fast.  However, design of a robot is simple,
the only thing you need to provide is the Denavit-Hartenberg matrices,
and a file describing what each link looks like (in the form: position
of the endpoints, and which endpoints are connected by a line).
The number of DoF is not a parameter, but requires recompilation.

And it is public domain, but I never put it on an ftp site since I was afraid
of getting too many bug reports which I then felt I had to honour.

So, if I get enough reactions I will tar the whole thing: documentation,
sources, the lot, and put it on an ftp site.

							Patrick

PS it has been compiled to run under SunOS and silicon graphics.
