Learning Distributed Controllers for V-Formation

TL;DR

This paper presents a method to synthesize a distributed neural controller for V-formation from a centralized MPC, demonstrating improved scalability and speed, while also establishing fundamental limitations of deterministic approaches.

Contribution

It introduces a novel deep learning approach with Counterexample-Guided Retraining to create scalable, distributed V-formation controllers, and proves the impossibility of certain deterministic solutions.

Findings

Neural controller generalizes from 7 to 15 agents.

Achieves substantial speedup over MPC-based controller.

Uses statistical model checking for confidence in convergence.

Abstract

We show how a high-performing, fully distributed and symmetric neural V-formation controller can be synthesized from a Centralized MPC (Model Predictive Control) controller using Deep Learning. This result is significant as we also establish that under very reasonable conditions, it is impossible to achieve V-formation using a deterministic, distributed, and symmetric controller. The learning process we use for the neural V-formation controller is significantly enhanced by CEGkR, a Counterexample-Guided k-fold Retraining technique we introduce, which extends prior work in this direction in important ways. Our experimental results show that our neural V-formation controller generalizes to a significantly larger number of agents than for which it was trained (from 7 to 15), and exhibits substantial speedup over the MPC-based controller. We use a form of statistical model checking to compute confidence intervals for our neural V-formation controller's convergence rate and time to convergence.