Pursuer Assignment and Control Strategies in Multi-Agent Pursuit-Evasion Under Uncertainties.

Area of Science:

• Robotics
• Artificial Intelligence
• Control Theory

Background:

• Pursuit-evasion problems are complex, especially with heterogeneous teams and uncertain agent states.
• Existing methods often lack efficient strategies for handling redundant pursuers or probabilistic information.

Purpose of the Study:

• To develop and evaluate an optimal assignment strategy for multiple pursuers in a pursuit-evasion scenario.
• To minimize the estimated capture time by effectively assigning pursuers, considering redundancy and probabilistic agent states.

Main Methods:

• Utilized Markov localization for probabilistic state estimation between pursuers and evaders.
• Developed a search-based control strategy for pursuers incorporating evader probability distributions.
• Employed a modified Hungarian algorithm and a novel redundant pursuer assignment algorithm.

Main Results:

• The proposed redundant pursuer assignment algorithm demonstrated superior performance compared to a nearest-neighbor approach.
• The strategy effectively minimizes estimated capture time in simulated pursuit-evasion scenarios.
• Evaders successfully predicted pursuer assignments and employed evasion tactics based on probabilistic knowledge.

Conclusions:

• The novel assignment strategy significantly enhances efficiency in multi-agent pursuit-evasion systems.
• Considering redundancy and probabilistic information is crucial for optimizing capture time.
• This approach offers a robust solution for complex heterogeneous pursuit-evasion games.