HSMS-Based Event-Triggered Adaptive Dynamic Programming for Pursuit-Evasion Differential Games of Multiagent Systems

Area of Science:

• Robotics and Control Systems
• Game Theory
• Artificial Intelligence

Background:

• Multiagent systems (MASs) face complex coordination challenges in pursuit-evasion differential games (PEDGs).
• Existing control strategies often struggle with communication overhead and suboptimal resource allocation.
• Ensuring cohesive team behavior while managing individual agent objectives is critical.

Purpose of the Study:

• To develop a distributed approximate optimal control scheme for multiagent pursuit-evasion games.
• To enhance coordination and communication efficiency in multiagent systems.
• To address the challenge of dynamic target allocation among pursuer agents.

Main Methods:

• Formulation of agent interactions using a divide-and-conquer algebraic graph approach.
• Implementation of a state event-triggered mechanism (ETM) for communication resource conservation.
• Design of a polymeric hierarchical sliding mode surface (HSMS) for improved system response.
• Development of a novel dynamic target allocation algorithm for rational pursuer assignment.
• Application of adaptive dynamic programming (ADP) with a single-critic neural network (NN) to solve Hamilton-Jacobi-Bellman (HJB) equations for optimal control policies.

Main Results:

• The proposed HSMS-based event-triggered optimal control policies effectively manage the PEDGs problem in MASs.
• The dynamic target allocation algorithm enhances team coordination among pursuers.
• The state event-triggered mechanism significantly conserves communication resources.
• Simulations demonstrate the effectiveness of the control scheme in a two-pursuer-two-evader scenario.

Conclusions:

• The integrated approach combining HSMS, ETM, and ADP provides a robust solution for distributed optimal control in PEDGs.
• The study validates the efficacy of the proposed methods in improving coordination, resource management, and overall system performance.
• This research contributes to the advancement of intelligent control strategies for complex multiagent systems.