Hi Ram, I don't have a complete answer for you, but the answer is that yes, we have done various simulations involving networks of robots.
We did a bunch of work involving Lego Mindstorms quite awhile ago. Most of this work did not ship in the release, but it was very cool. We also did some work involving controlling a robot with a ducted fan. http://sec.eecs.berkeley.edu/reports/03/1 says: > Steve Neuendorffer tested his "crazyboard" controller models, built in > Ptolemy II, on the Caltech hardware. There was one laptop sitting on > the Crazyboard and running a very simple model to receive control > commands, and there was another laptop that runs the control algorithm > remotely and sends commands to the laptop on the Crazyboard. The > reason for two laptops is that they wanted to test different control > algorithms and modify algorithms conveniently. But this approach has a > drawback: the latency is quite big due to the control being over the > network. > > Yang and Steve have suggested an alternative architecture where the > control law is implemented in a migrating model that is transported > over the network and executed locally. Yang built an experimental > setup where a MobileModel actor accepts a model over the network and > executes that model to process streaming inputs. The model can be > updated in real time by a supervisory model on the second laptop, but > the control loop no longer includes the network, thus greatly > improving performance. This work continues, with the objective of > building an infrastructure that is easy for control systems designers > to use. > > Yang has also experimented with using JXTA to discover control > services and Crazyboard configurations. JXTA is a discovery > infrastructure from Sun Microsystems. > > A key observation that Yang and Steve made is that there is a need for > a modeling and simulation environment for developers designing > distributed systems based on OCP. Designers today can use Simulink to > draw the component diagram of a system, but they cannot do the > simulation in Simulink since it does not provide the proper semantics. > It would be possible to build a Ptolemy II domain with OCP semantics > that might serve this purpose. More recently, the wireless domain has been used to simulate evader/pursuer games and setting up ad hoc networks. Many of the wireless demos would be of interest to you. They can be found in the release at ptII4.0.1/ptolemy/domains/wireless/demo or on line at http://ptolemy.eecs.berkeley.edu/ptolemyII/ptII4.0/ptII4.0.1/ptolemy/domains/wireless/doc/ You should also check out the continuous time CarTracking demo. The upcoming release will have a continuous time Vehicle to Vehicle demo as well. Ptolemy II 5.0-beta will be released next week. The Ptolemy miniconference is next week in Berkeley, let me know if you want to attend. _Christopher Christopher Brooks (cxh at eecs berkeley edu) University of California Programmer/Analyst Chess/Ptolemy/Trust US Mail: 558 Cory Hall #1770 ph: 510.643.9841 fax:510.642.2739 Berkeley, CA 94720-1770 home: (F-Tu) 707.665.0131 (W-F) 510.655.5480 (office: 400A Cory) -------- Hi, I am new to the Ptolemy framework and I am curious to know if my particular simulation requirements can be handled by it. Specifically, I am interested in modeling control and coordination algorithms for a team of robots over a network. The motion of the robots is modeled using continuous time differential equations. The communication between the robots takes place over an ad-hoc network. Due to the motion of the robots, the ad-hoc network topology is continuously changing. Are there any specific examples that I can look at for understanding how to do this ? Thanks, -Ram -------- ---------------------------------------------------------------------------- Posted to the ptolemy-hackers mailing list. Please send administrative mail for this list to: [EMAIL PROTECTED]
