Game-Based Approximate Optimal Motion Planning for Safe Human-Swarm Interaction.

Area of Science:

• Robotics
• Control Theory
• Game Theory

Background:

• Human-swarm interaction safety is crucial and challenging.
• Existing methods require high real-time computation for constrained optimization.
• Need for robust and autonomous safety solutions in human-robot systems.

Purpose of the Study:

• To develop a safe and robust framework for human-swarm interaction.
• To address real-time limitations of current safety approaches.
• To enable autonomous trajectory modification for unsafe human commands.

Main Methods:

• Formulating the problem as a Stackelberg-Nash game over the entire time domain.
• Designing best-response controllers (Nash equilibrium) for follower robots.
• Developing a Lyapunov-like control barrier function for leader safety and a learning-based controller for formation tracking.

Main Results:

• Robotic swarms successfully maintain desired formations while tracking human commands.
• Controllers autonomously modify trajectories to ensure safety against unsafe commands.
• Safety is maintained even with dynamic obstacles present, verified by simulations and experiments.

Conclusions:

• The proposed Stackelberg-Nash game approach effectively ensures safety in human-swarm interaction.
• The method overcomes real-time constraints by optimizing over the entire time domain.
• The system demonstrates robust cooperative behavior and autonomous safety adjustments.