A Distributed, Event-Triggered, Adaptive Controller for Cooperative Manipulation With Rolling Contacts

Abstract

We present a distributed, event-triggered, and adaptive control algorithm for cooperative object manipulation with rolling contacts and unknown dynamic parameters. Whereas conventional cooperative manipulation methods require rigid contact points, our approach exploits rolling effects of passive end-effectors and does not require force/torque sensing. The removal of rigidity allows for more modular grasping, increased application to more object types, and online adjustment of the grasp. The proposed control algorithm exhibits the following properties: 1) it is distributed, in the sense that the robotic agents calculate their own control signal, under an event-triggered communication scheme. Such a scheme reduces the interagent communication requirements with respect to continuous communication schemes; 2) it uses an online adaptation mechanism to accommodate for unknown dynamic parameters of the object and the agents and 3) it adapts existing internal force controllers to guarantee no slip throughout the manipulation task despite the event-triggered nature of the communication scheme. Hardware implementation validates the effectiveness of the proposed approach.