|
Our goal in this project is to develop algorithms to search
for targets in complex environments. Our techniques decouple
the actions of the pursuers and employ a limited horizon search
with a cost heuristic. This allows for the following virtues:
- Scalability to multiple robots and large environments
- Incoporation of motion models of targets and prior information on target's state
- Utilization of information from noisy measurements
We applied our algorithms to three sub-problems in the coordinated search domain.
For each of these sub-problems, we tested in three real-world environments. The sub-problems are described below:
- Efficient search: find a target with a given motion model in the lowest expected time
- Guaranteed search: clear the environment so that capture is guaranteed
- Constrained search: maintain inter-robot constraints such as network connectivity or line-of-sight communication
| Coordination Problems |
Efficient Search |
Guaranteed Search |
Constrained Search |
| House |
|
|
|
| Office |
|
|
|
| Museum |
|
|
|
Performance Results for Coordinated Search in Different Environments
|
