Distributed Heterogeneous Multi-Agent Networks Optimization with Nonconvex Velocity Constraints.

• Distributed heterogeneous multi-agent optimization is investigated. • The nonconvex velocity constraints are considered in this paper. • An improved distributed algorithm is introduced. • A time varying transformation is employed. • The convergence analysis of the algorithm is completed with the aid of some new Lyapunov function. This paper briefly deals with the distributed optimization problem over heterogeneous multi-agent networks with nonconvex velocity constraints. Based on the constraint operator, a distributed optimization algorithm is proposed for each agent, where the gradient gains are nonuniform and only local neighbors' information is used by each agent. Then, a time-varying coordination transformation is introduced to change the closed-loop system into a new system, which is composed of linear part and nonlinear part. It is proved that the system matrices of linear part are time-varying stochastic matrices and the system states are bounded. Furthermore, under some mild assumptions, it is shown that all agents' position states could converge to the optimal solution of the team objective function, while the velocities of all agents are constrained to the corresponding nonconvex sets. Finally, the feasibility of the proposed algorithm is verified by simulations. [ABSTRACT FROM AUTHOR]