Hierarchical distributed receding horizon control for a group of agents

This paper is concerned with a hierarchical distributed receding horizon control (HDRHC) approach, by which a global objective of locating the peaks of an unknown environment of interest can be achieved among locally communicating agents. The proposed HDRHC approach is executed by each agent independently and consists of two levels. In the first level, a radial basis function network is used to model the unknown environment of interest. On the basis of the established environment model, a dynamical optimization problem is formulated and solved by using a receding horizon control approach such that an ideal movement trajectory for each agent is generated. The agents can trace the peaks of the environment of interest by moving along the ideal movement trajectory; however, the collision among agents may occur. In the second level, a cooperative control optimization problem, whose aim is to avoid collision among agents, is designed. Hence, the real movement trajectory of each agent, which is produced by using the receding horizon control approach, not only should minimize the cooperative control optimization problem, but also should be close to the ideal movement trajectory. Finally, the effectiveness of the proposed HDRHC approach is illustrated for the gradient climbing problem.