Passivity-Based Distributed Control for Cooperative Unknown Payload Manipulation with Sensing/Actuation Delays

We consider the problem of remotely manipulating an unknown payload using multiple agents in the presence of sensing and actuation delays. The overall objective is to find an optimal input force for each agent so that the linear and angular velocities of a rigid object track a reference. We prove that combining primal-dual gradient dynamics with a passivity-based technique called the scattering transformation yields a completely distributed algorithm that converges locally to the optimal solution for the associated optimization program when the delays are homogeneous. We also show that when the delays are heterogeneous, the algorithm is still guaranteed to converge, but to a possibly suboptimal solution. Numerical simulations demonstrate the velocity tracking performance and optimality of the resulting force inputs on practical problems.