Cooperative Approximate Optimal Indirect Regulation of Uncooperative Agents with Lyapunov-Based Deep Neural Network
Wanjiku A. Makumi, Zachary I. Bell, Jhyv N. Philor, Warren E. Dixon · 2024
An approximate optimal indirect regulation problem is considered that consists of cooperative herding agents and a target agent coupled by uncertain interaction dynamics. Approximate dynamic programming is used to control the cooperative herding agents to optimally influence the target agent to a goal location while learning the optimal formation of the agents to achieve this task. Since the interaction dynamic between the agents is unknown, a Lyapunov-based deep neural network is used with an integral concurrent learning-based adaptive update law to facilitate system identification. A Lyapunov-based stability analysis is used to show uniformly ultimately bounded stability of the system.