| Sponsor | Prof. Sven Koenig <skoenig@cc.gatech.edu> |
| Area | Intelligent Systems |
Explore how several robots can map an environment cooperatively. You might want to start with the following simple strategy: The robots share the map and each robot always moves to the closest location that no robot has visited yet. Initially, assume that the robots operate in a simple (simulated) gridworld without actuator and sensor uncertainty and implement this algorithm. Determine experimentally how the time and the total travel distance depend on the number of robots and frequency of communication. (If you want to do a really great job, think about whether it is a good idea to use D* as part of the implementation.) Then, change the algorithm so that it can cope with actuator and sensor uncertainty.
Optimal and Efficient Path Planning for Partially-Known Environments. Stentz, ICRA 94.
The Focussed D* Algorithm for Real-Time Replanning. Stentz, IJCAI 95.