Deep Reinforcement Learning Framework for Coordinated Multi-Robot Flocking

Multi-robot flocking systems' robustness is bad, adaptability is week, and communication efficiency is low in complex and dynamic environments. Traditional control system has high risk and high cost.

• Leveraging advanced deep reinforcement learning (DRL) methodologies, this technology combines asymmetric self-play mechanisms with sophisticated communication refinement and enhancement techniques. By training flocking robots alongside learnable adversarial interferers, the system fosters resilient flocking strategies.

• It significantly enhance multi-robot flocking systems' robustness, adaptability, and communication efficiency in complex and dynamic environments.
• It optimizes inter-robot communication, ensuring selective interaction and consensus-driven information exchange.
• This technology not only mitigates operational risks and costs but also overcomes the limitations of traditional control systems.

• Gold Medal at the 4th Asia Exhibition of Innovations and Inventions (AEII)

• Urban Drone Delivery: Enhancing the safety and efficiency of drone swarms for package delivery in densely populated urban areas.
• Logistics and Supply Chain Management: Streamlining warehouse operations and automated inventory management through robust multi-robot systems.
• Search and Rescue Operations: Providing resilient and adaptable robotic teams capable of navigating and operating in disaster-stricken or hazardous environments.
• Mobile Surveillance and Security: Deploying coordinated robot swarms for comprehensive monitoring and security tasks in dynamic and obstacle-rich settings.