Co-Manipulation in Leader-Follower

 For tasks involving large or heavy objects that exceed the capability of a single robot, centralized multi-robot systems can enable effective collaboration. However, in environments with limited communication or sensing, sharing target trajectory or robot states becomes challenging, requiring additional collaboration strategies. In systems where only the leader robot has task information, achieving coordinated action becomes difficult for follower robots. To address this, it is essential to estimate unknown parameters, including object dynamics and kinematic relationships between the object and the robots. Based on such estimations, collaboration should aim not merely to follow but to assist actively by enhancing the leader's task efficiency. Accurate estimation of object parameters allows for optimal workload allocation, enabling efficient collaboration between leader and follower robots without relying on direct communication.