Hybrid ellipsoidal learning and fuzzy control for platoons of smart cars

A fuzzy system controls gaps between cars in single lane platoons. Fuzzy controllers create, maintain, and divide platoons on the highway. Each car´s controller uses only data from sensors on the car. Tightly coupled platoons avoid the “slinky effect“ by dropping back during platoon maneuvers. When the lead car reaches its goal, the follower cars return to the proper platoon spacing. Differences in car and engine types require changes in fuzzy rules and sets. A hybrid neural-fuzzy system combines supervised and unsupervised learning to find and tune the fuzzy-rules. Unsupervised competitive learning chooses the first set of ellipsoidal fuzzy rules. Supervised learning tunes the fuzzy rules with gradient descent. The authors tested the fuzzy gap controller with a realistic car model