The objective of this thesis is to develop a strategy that allows robots to safely execute open-loop motion patterns for pre-computed time durations when facing interruptions in
communication. By computing the time horizon in which collisions with other robots are impossible, this method allows the robots to move safely despite having no updated
information about the environment. As the complexity of multi-robot systems increase, communication failures in the form of packet losses, saturated network channels and hardware failures are inevitable. This thesis is motivated by the need to increase the robustness of operation in the face of such failures. The advantage of this strategy is that it prevents the jerky and unpredictable motion behaviour which often plague robotic systems experiencing communication issues. To compute the safe time horizon, the first step involves constructing reachable sets around the robots to determine the set of all positions that can be reached by the robot in a given amount of time. In order to avoid complications arising from the non-convexity of these reachable sets, analytical expressions for minimum area ellipses enclosing the reachable sets are obtained. By using a fast gradient descent based technique, intersections are computed between a robot’s trajectory and the reachable sets of other robots. This information is then used to compute the safe time horizon for each robot in real time. To this end, provable safety guarantees are formulated to ensure
collision avoidance. This strategy has been verified in simulation as well as on a team of two-wheeled differential drive robots on a multi-robot testbed.
| Identifer | oai:union.ndltd.org:GATECH/oai:smartech.gatech.edu:1853/55067 |
| Date | 27 May 2016 |
| Creators | Mayya, Siddharth |
| Contributors | Egerstedt, Magnus |
| Publisher | Georgia Institute of Technology |
| Source Sets | Georgia Tech Electronic Thesis and Dissertation Archive |
| Language | en_US |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
Page generated in 0.0017 seconds