A method of combining BP neural network with PID controller is applied to two manipulators cooperating a rigid payload. Based on kinetic constraints of two robots manipulating one rigid object

different control subsystems are designed for the leader and follower according to the roles they play in the system. PID parameters can be rectified online in the two subsystems

which makes the parameters achieve optimized values. With the optimized parameters

the control system can be more robust and more adaptive. The validity of the adoption was verified by numerical simulation.