Coordinating Reactive Behaviors

Combinating reactivity with planning has been proposed as a means of compensating for potentially slow response times of planners while still making progress toward long term goals. The demands of rapid response and the complexity of many environments make it diicult to decompose, tune and coordinate reactive behaviors while ensuring consistency. Neural networks can address the tuning problem, but are less useful for decomposition and coordination. We hypothesize that interacting reactions can be decomposed into separate behaviors resident in separate networks and that the interaction can be coordinated through the tuning mechanism and a higher level controller. To explore these issues, we have implemented a neural network architecture as the reactive component of a two layer control system for a simulated race car. By varying the architecture, we test whether decomposing reactivity into separate behaviors leads to superior overall performance, coordination and learning convergence.