Project 1: Distributed Sensing, Learning and Control in Mobile Sensor Networks

1SealDOD01.jpg.opt186x185o0,0s186x185

(Supported by the US Department of Defense)

Authors: Hung M. La, Ronny S. Lim and Weihua Sheng

Introduction:

Sensor network, especially mobile sensor network (MSN) has been extensively studied in recent years. An MSN has some advantages over a stationary sensor network (SSN) such as: the adaptation to environmental changes and the reconfigurability for better sensing performance. Therefore, MSN can be applied in many applications such as target tracking and protection of endangered of species. Figure 1 shows the example of target surveillance

Goal:

We are developing a framework for cooperative sensing, learning and control in distributed fashion for mobile sensor networks. Our goal is to design a control law to run on each mobile sensor platform to obtain the desired collective behavior. Example tasks include target(s) tracking in noisy or noisy-free environments, formation control, multi-agent learning to avoid predators, distributed sensor fusion for scalar field mapping.  Our constraints are that each sensor platform may have limited computational power, sensing and communication capabilities.

This research can benefit battlefield surveillance, security monitoring.

Publications of this project

  • Hung M. La, and Weihua Sheng, Flocking control algorithms for multiple agents in cluttered and noisy environments, Bio-Inspired Self-Organizing Robotic Systems,Studies in Computational Intelligence, Springer-Verlag Berlin Heidelberg, 2011, Vol. 355, pp. 53-79, DOI:10.1007/978-3-642-20760-0_
  • Hung M. La and Weihua Sheng, Dynamic targets tracking and observing in a mobile sensor network, the Elsevier journal on Robotics and Autonomous Systems, 2011.
  • Weihua Sheng and Hung M. La, Network of cooperating mobile sensors used for mapping, SPIE Newsroom / Defense & Security, August, 2011.
  • Hung M. La and Weihua Sheng, Cooperative sensing in mobile sensor networks based on distributed consensus, the Signal and Data Processing of Small Targets conference, Proceedings of SPIE, August 23 – 25, 2011, San Diego, Californima, USA.
  • Hung M. La, Ronny S. Lim, Heping Chen and Weihua Sheng, Decentralized flocking control with minority of informed agents, in the proceedings of the IEEE Conference on Industrial Electronics and Applications (ICIEA), June 21-23, 2011, Beijing, China.
  • Hung M. La, Ronny S. Lim and Weihua Sheng, Hybrid system of reinforcement learning and flocking control in multi-robot domain, in the proceedings of the Conference on Theoretical and Applied Computer Science (TACS), November 5, 2010, Stillwater, Oklahoma, USA. Best Paper Award.
  • Hung M. La and Weihua Sheng, Flocking control of multiple agents in noisy environments. in the proceedings of the IEEE International Conference on Robotic and Automation (ICRA), May 3-8, 2010, Alaska, USA.
  • Hung M. La and Weihua Sheng, Multi-target tracking and observing in mobile sensor networks, in the proceedings of the Conference on Theoretical and Applied Computer Science (TACS09), October 24th, 2009, Stillwater, Oklahoma, USA.Best Paper Award.
  • Hung M. La and Weihua Sheng, Adaptive flocking control for dynamic target tracking in mobile sensor networks, in the proceedings of the 2009 IEEE International Conference on Intelligent Robots and Systems (IROS), October 11 – 15, 2009, St. Louis, Missouri, USA.
  • Hung M. La and Weihua Sheng, Flocking control of a mobile sensor network to track and observe a moving target, inthe proceedings of the IEEE International Conference on Robotics and Automation (ICRA), May 12 – 17, 2009, Kobe, Japan.
  • Hung M. La and Weihua Sheng, Moving targets tracking and observing in a distributed mobile sensor network, in the proceedings of the 2009 American Control Conference (ACC), June 10 – 12, 2009, St. Louis, Missouri, USA. Best Presentation of Session.
  • Hung M. La and Weihua Sheng, Flocking Control Algorithms for Multiple Agents in Cluttered and Noisy Environments, Workshop in Bio-Inspired Self-Organizing Robotic Systems on the IEEE International Conference on Robotics and Automation (ICRA), May 3 – 8, 2010, Alaska, USA.

 

battle field

Figure 1. Application of MSN in the battle field (Murray 2007).

fish_school2

Figure 2. School of fish.

gradient_result1

This figure shows 20 agents climbing on the top of the interested field based its own gradient.

 

This video is for  adaptive flocking control algorithm. The network of mobile sensors will shrink its size automatically in distributed fashion in order to pass through the narrow space between the obstacles (red elipses). The goal is to maintain the formation and connectivity of the network. Please see published paper, “Adaptive flocking control for dynamic target tracking in mobile sensor network”, IROS 2009,  for more details.

 

This video is for adaptive flocking control algorithm. The network of 7 Rovio robots will shrink its size automatically in distributed fashion in order to pass through the narrow space between the obstacles (boxes). The goal is to maintain the formation and connectivity of the network. For more details please see Flocking control of multiple agents in cluttered and noisy environments,   in Bio-Inspired Self-Organizing Multi-Agent Systems, Studies on Complexity Intelligence Book Series, Springer.

 

This video is for the paper “Decentralized Flocking Control with Minority of Informed Agents”  ICIEA2011. (Experiment 7 Rovio robots )

This video is for the paper “Decentralized Flocking Control with Minority of Informed Agents”  ICIEA2011. (Simulation with 50 robots)

 

This video is for Reinforcement Learning of Cooperative Behaviors in Multi-robot Flocking to avoid predator”.

This video shows 7 robots flocking together to do cooperative sensing. The scalar field is modeled by multiple cells. Each robot measures the value at cells in its sensing range (blue circle) and cooperate with its neighbors to build the whole map of the scalar field.

This video shows the Multi-CoM-Cohesion algorithm of multi-agent cooperation in noisy environment. The connectivity is maintained here.

This video shown multi-target tracking in a mobile sensor network. Initially, a group of robots flock together based on flocking control algorithm to track a target moving in sine wave (red line). When the second target appears hafl of robots will split to track this target,…For more information please see the published paper “Moving targets tracking and observing in a distributed mobile sensor network“, ACC 2009.

 

This video shows the optimal flocking control algorithm where the flocking parameters are selected based on Genetic Algorithm.