Cooperative control of payload transport by mobile manipulator collectives
Multi-manipulators based mobile manipulation is an important capability to extend the domain of robotic applications. The novel feature endowed by the combination of mobility with manipulation is crucial for a number of applications, ranging from material handling task to planetary exploration. The benefits include increased workspace, reconfigurability, improved disturbance rejection capabilities and robustness to failure. The challenges, however, arise from the compatibility of various holonomic and nonholonomic constraints and kinematic and dynamic redundancy. Moreover, cooperative manipulation would lead to significant dynamic coupling and requires delicate motion coordination. Failure to consider these effects can cause excessive internal forces and high energy consumption, and even destabilize the system. To deal with these entailed issues, we present a decentralized dynamic control algorithm for a robot collective consisting of multiple nonholonomic wheeled mobile manipulators capable of cooperatively transporting a common payload. The nonholonomic wheeled mobile manipulator consists of a fully-actuated manipulator arm mounted on a disk-wheeled mobile base. In this algorithm, the high level controller deals with motion/force control of the payload, at the same time distributes the motion/force task into individual agents by grasp description matrix. At each individual agent, the low level controller decomposes the system dynamics into decoupled task space (end-effector motions/forces) and a dynamically-consistent null-space (internal motions/forces) component. The agent level control algorithm facilitates the prioritized operational task accomplishment with the end-effector impedance-mode controller and secondary null-space control. The scalability and modularity is guaranteed upon the decentralized control architecture. Within the dynamic redundancy resolution framework, a decentralized coordination and control with collision avoidance capability is further studied for mobile manipulator collectives. A variety of numerical simulations are performed for multiple mobile manipulator system carrying a payload (with/without uncertainty) to validate this approach. The simulations test the capability of internal force regulation by cooperative manipulators. The end-effector and mobile base tracking capability is also verified in the simulations. Multiple mobile manipulator collision avoidance is also studied in simulation. A force control system with end-effector and force sensor is designed and some basic force control tests are performed by this platform.