LEMURS: Learning Distributed Multi-Robot Interactions
Eduardo Sebastián,Thai Duong,Nikolay Atanasov,Eduardo Montijano,Carlos Sagüés,Eduardo Sebastián,Thai Duong,Nikolay Atanasov,Eduardo Montijano,Carlos Sagüés
This paper presents LEMURS, an algorithm for learning scalable multi-robot control policies from cooperative task demonstrations. We propose a port-Hamiltonian description of the multi-robot system to exploit universal physical constraints in interconnected systems and achieve closed-loop stability. We represent a multi-robot control policy using an architecture that combines self-attention mechan...
