Task-oriented control and coordination of multi-agent systems under varying constraints
Time: Thu 2021-02-11 15.00
Location: Q2, Malvinas väg 10, Stockholm (English)
Subject area: Electrical Engineering
Doctoral student: Pian Yu , Reglerteknik
Opponent: Professor Christos Cassandras, Boston University
Supervisor: Dimos V. Dimarogonas, Reglerteknik, Centrum för autonoma system, CAS, ACCESS Linnaeus Centre; Carlo Fischione, Nätverk och systemteknik, ACCESS Linnaeus Centre
Multi-agent systems (MAS) offer a tremendous potential to improve the quality of modern society life. For instance, robot networks have been widely used for providing services such as search and rescue missions, surveillance and data collection, healthcare and entertainment. One challenge for MAS is the design of control and coordination strategies that enable each agent to perform operations safely and efficiently while achieving group or individual motion objectives. This dissertation addresses this challenge by developing different task-oriented control and coordination strategies for MAS.
The first part of the thesis is devoted to the control of MAS under cooperative tasks. Firstly, we investigate the distributed control of multi-agent consensus. Event-triggered control strategies are developed to reduce communication burden. It is proven that consensus can be achieved with the proposed strategies. Next, we tackle the problem of leader-follower multi-agent target tracking, where the group of agents is assigned a set of dynamically activated tasks, each of which has to be completed within a deadline. A dynamic scheduling strategy is proposed and distributed control laws are designed respectively for the leader and follower agents. It is shown that the proposed control laws guarantee the satisfaction of each task.
In the second part of the thesis, we develop control and coordination schemes for single- or multi-agent systems under temporal logic specifications. Firstly, the symbolic control of continuous-time uncertain nonlinear systems is studied. A new stability notion called approximate controlled globally practically stable is introduced. Building on this notion, we provide for the first time a behavioral relationship between the original continuous-time system and its discrete state-space symbolic model. After that, we consider the robust satisfiability check and online control synthesis of uncertain systems under signal temporal logic specifications. A sufficient condition is obtained for the robust satisfiability check of the uncertain systems. An online control synthesis algorithm is designed, which is shown to be sound for uncertain systems and both sound and complete for deterministic systems. Finally, the motion coordination of MAS is investigated, where each agent is assigned a linear temporal logic specification. Based on the realistic assumptions that each agent is subject to both state and input constraints and can have only local view and local information, a provably safe and fully distributed multi-agent motion coordination strategy is proposed.