Networked Control of Autonomous Systems

Project Objectives: By unifying techniques from control theory, graph theory, communication, and estimation technology, the research aims to systematically design networked control strategies for autonomous assets to perform cooperative tasks (e.g., navigation, surveillance, etc.) in a complex environment with various constraints, such as network connectivity constraints, sensor constraints, and communication bandwidth constraints.

 

  • Cooperative Control over Random Communication Graph

Abstract: Agents can experience unexpected loss of communication with other team members in a complex environment (e.g., urban environment), resulting in a disconnected network and mission failure. To overcome the challenge of random communication, I have developed decentralized controllers for networked agents to only use intermittent information from neighboring agents to

  • achieve simultaneous arrival at a predetermined target in [1] while restricting the motion of the group within a desired region and evenly spacing the agents' orientation,
  • perform cooperative target tracking in [2] with random failures in observing the target while preserving the connectivity of the position dependent network, 
  • establish almost sure convergence result in [3] for the followers' states to the convex hull spanned by the leaders' states over directed leaders' states over directed leader-follower random networks.

Related Publications: 

[1]. Z. Kan, S. S. Mehta, J. M. Shea, J. W. Curtis, and W.E. Dixon, "Balanced Containment Control and Cooperative Timing of a Multi-Agent System over Random Communication Graphs," Automactica, under review.
[2]. Z. Kan, S. S. Mehta, T. Cheng, and W. E. Dixon, "Target Tracking with Distributed Control of Algebraic Connectivity over Random Communication Networks," International Journal of Robust and Nonlinear Control, under review.
[3]. Z. Kan, J. M. Shea, and W. E. Dixon, "Leader-Followers Containment Control Over Directed Random Graphs," Automatica, under review

 

  • Graph Matching based Formation Control

​Abstract​tGreat efficiency and operational capability can be realized by networked agents in particular formations. To perform formation reconfiguration with minimum amount of agent movement for a group of identical agents with limited communication capabilities, based on graph matching algorithms, I have developed 

  • a node mapping strategy to specify the node correspondence between an arbitrary initial graph and the desired graph
  • developed a decentralized control scheme to ensure convergence of all agents to the desired configuration with collision avoidance among agents and preservation of network connectivity

Related Publications:

[4]. Z. Kan, L. Navaravong, J.M. Shea, E. L. Pasiliao, and W. E. Dixon, "Graph Matching Based Formation Reconfiguration of Networked Agents with Connectivity Maintenance, "IEEE Transactions on Control of Network Systems, Vol. 2 No 1., pp 24-35 (2015).

 

  • Vision-based Missile Control

Abstract: I have collaborated with researchers from AFRL and UF to develop a vision-based approach in [5] for relative navigation and robust control of a missile to intercept a ground target maneuvering with unknown time-varying velocity. The developed approach enables design of a robust and adaptive nonlinear visual servo controller to compensate for the unknown time-varying missile-target relative velocity range as well as nonlinear bounded exogenous disturbances and parametric uncertainties.

Related Publications:

[5]. S. S. Mehta, Z. Kan, E. L. Pasiliao, J. W. Curtis, "Vision-Based Relative Navigation and Control of a Sensorless Missile," Journal of the Franklin Institute, under review

 

  • Event-triggered Hybrid Control

Abstract: Continuous information exchange between agents through communication is not desirable for certain networks with limited bandwidth constrains. To reduce the frequency of inter-agent communication, I have collaborated with colleagues at UF in developing a hybrid decentralized event-triggered control scheme in  [6] to achieve consensus for a leader-follower network by using less inter-agent communication. In contrast to the classical periodic sampling, the developed hybrid controller only triggers inter-agent communication when the error indicator grows beyond a threshold, and achieves the same level of performance using lower sampling frequency. 

Related Publications:

[6]. T. Cheng, Z. Kan, J. M. Shea, and W. E. Dixon, "Decentralized Event-triggered Control for Leader-follower Consensus, "IEEE Conference on Decision and Control, Los Angeles, CA, 2014, pp.1244-1249.