A Leader-Selection Approach to Fuzzy Formation Control of Autonomous Vehicles
Open Access
- Author:
- Thomson, Brian Francis
- Graduate Program:
- Electrical Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- July 13, 2009
- Committee Members:
- Jeffrey Scott Mayer, Thesis Advisor/Co-Advisor
Randy Young, Thesis Advisor/Co-Advisor - Keywords:
- autonomous vehicles
leader selection
fuzzy logic
formation control - Abstract:
- The use of autonomous vehicles in a variety of applications has recently increased due largely to their ability to perform objectives too difficult for manned platforms. The cooperation between these vehicles can facilitate achievement of objectives. Formation control is described here as controlling the position alignment among a group of autonomous vehicles. Formation control of multiple autonomous vehicles is particularly useful for environment searching, exploring, and mapping. Vehicles can search a given environment and adapt the formation as more information becomes available about the objectives and the environment. Further, the vehicles can explore potentially hazardous or remote areas to develop a map of the area. The vehicles can improve their total search area by aligning themselves in certain formations based on sensor limitations. This thesis shows that a collection of fuzzy controllers, or fuzzy inference systems (FIS), can provide formation control of multiple homogeneous autonomous vehicles. A leader-follower approach is presented wherein the leader represents the reference vehicle for the formation, and the followers comprise the other vehicles in the group. Each vehicle is given an arbitrary initial direction and position in two dimensional spaces. A leader selection FIS determines the leader vehicle based on its direction and location relative to the desired formation line-of-travel. Once the leader vehicle is chosen, the objective is for each follower vehicle to converge to a desired position relative to the leader. Under the assumption that the follower vehicles can perfectly sense the states (i.e. velocity, position for the scope of this problem) of the leader vehicle and that the leader has been selected, the fuzzy controller generates velocity commands for each follower while the leader traverses an arbitrary path. The advantage of this approach is that fuzzy controllers can be tuned for any set of possible vehicle states. Thus only the set of possible vehicle states are required to develop a comprehensive formation controller.