Coordinated Control of Multi-agent Systems With Real-World Constraints
Abstract
In a battle field application, multi-agent systems are expected to be tough in a harsh environment, aggressive in pursuing optimal operation, yet efficient in expending resources. Such expectations to a large degree translate into requirements on the coordinated control algorithms, including their resiliency against disturbances and latencies of the communication topologies, their full utilization of the actuator capacities to achieve maximum performance, their efficiency in utilizing on-board resources, and their ability to maintain and build connectivity. The many major breakthroughs in the study of multi-agent systems have paved the foundation for the analysis and design of coordinated control systems in the presence of real-world constraints, which are practically imperative and theoretically challenging. The proposed research aims to develop coordinated control design tools for multi-agent systems operating in real-world situations and with real-world constraints. Four aspects of coordinated control design will be addressed: 1) coordinated control in the presence of actuator saturation; 2) coordinated disturbance tolerance and rejection; 3) connectivity building; and 4) coordinated control in the presence of time delays. Thrust 1: Coordinated control in the presence of actuator saturation. Actuator saturation degrades control system performance and may even cause instability. Saturation avoidance design leads to under-utilization of the actuator capacity and hence less than achievable performance. We will develop distributed control algorithms that allow the actuators to work in their saturation mode for better performance. We will also develop distributed optimal control algorithms which cause the agents to operate at a state that optimizes an objective function of the overall multi-agent system. Event-triggered control design will also be explored to implement the distributed control algorithms without requiring continuous updating of the control signals. Thrust 2: Coordinated disturbance rejection. Disturbances are inevitable in a real-world environment. Beside the disturbances external to the multi-agent systems (wind gusts or ocean currents), in a tight formation of unmanned aerial vehicles or unmanned underwater vehicles, disturbing to the surrounding flows by the neighboring vehicles affects the operation of a vehicle. We will develop coordinated control algorithms that maximize a multi-agent systemÕs capability to perform the desired tasks in the presence of disturbances. Thrust 3: Connectivity building. A state-dependent interaction topology is often the case in real world applications. Two agents can exchange information only when they are within a certain distance. Existing results on the coordinated control require such a state-dependent topology, which may be switching, to remain connected all the time. We will explore the interdependency between coordinated control and connectivity. Connectivity enables coordinated control, while coordinated control builds connectivity. We will develop coordinated control algorithms that build connectivity instead of requiring connectivity all the time, including the initial time. Thrust 4: Coordinated control in the presence of time delays. Because of the finite speeds of transmission and spreading as well as traffic congestions, there usually exist time delays in spreading and communication in reality. Much of the literature on multi-agent systems with time delays focuses on analyzing the ability of an existing control algorithm to tolerate time delays. We will take time delays into consideration in the design of coordinated control algorithms.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 11, 2018
- Source ID
- W911NF1710535
Entities
People
- Zongli Lin
Organizations
- Army Contracting Command
- United States Army
- University of Virginia