This paper investigates a class of multiple redundant manipulators collaborative operation with comprehensive constraint tasks from the perspective of aggregative game, and the aggregated information in the game model is not accessible to the participants (players). We incorporate each participant’s estimation of the aggregated information to guide the creation of the algorithm. Since each manipulator only transmits the aggregated kinematic information, this aggregative game model reduces the communication cost of multiple manipulators network. The proposed neurodynamic algorithm is capable of handling tasks such as trajectory tracking, obstacle avoidance and posture maintenance, and can also optimize the rarely-discussed self-adjustment. In view of the unknowability of the kinematics of the manipulator, we introduce a neural operator based on fuzzy association rules to approximate the error. The convergence of the neurodynamic algorithm for generalized Nash equilibrium (GNE) is provable, and it can achieve distributed optimal control of the multiple manipulators.



