Distributed Perception and Control of Multi-Robot Systems
Teacher
Eduardo Montijano
Eduardo Montijano is an assistant professor at Universidad de Zaragoza, Spain. He received the M.Sc. and Ph.D. degrees from the Universidad de Zaragoza, Spain, in 2008 and 2012 respectively, obtaining the extraordinary award of the Universidad de Zaragoza in the 2012–2013 academic year. He has been a visiting scholar at University of California San Diego, University of California Berkeley and Boston University in the United States and at Royal Institute of Technology, in Stockholm, Sweden. He has also been a faculty member at Centro Universitario de la Defensa, Zaragoza, Spain, between 2012 and 2016. His main research interests include distributed algorithms, cooperative control and computer vision.
Abstract
This course will be devoted to the problems of distributed perception and control with teams of mobile robots. A consistent perception of the environment is crucial for the good development of any multirobot application. In order to make this possible, the robots need to communicate with each other and fuse joint observations. However, in a realistic scenario, distributed solutions to this problem are not trivial. In the first part of the course, we will present a deep study of distributed consensus algorithms and how a team of robots equipped can use them for cooperative perception with vision. We will address several important issues, such as bandwidth limitations, dynamic features and outlier information, which appear when using these sensors. In the second part of the course, we will pay attention to problems related to the control of the team of robots. On one hand, we will discuss how to reach flexible formations in an optimal manner. On the other hand, we will present coordination strategies for the problem of persistent monitoring using a team of mobile robots.
Program
- Lesson one (2nd October 15:00-17:00). Fast Distributed Consensus in Robotic Networks
- The consensus problem
- Speeding up linear iterations using Chebyshev polynomials
- Lesson two (4th October 16:00-18:00). Discrete time dynamic average consensus
- Dynamic average consensus
- Robustness to initialization errors
- Lesson three (5th October 15:00-17:00) Laboratory: Evaluation of the algorithms in Matlab
- Lesson four (9th October 15:00-17:00). Distributed Data Association and Robustness to Outliers
- Distributed data association of multiple features
- Distributed consensus with robustness to outliers
- Lesson five (11th October 16:00-18:00). Distributed formation control of freely reachable formations
- Consensus-based distributed formation control of fixed formations
- Optimal freely reachable formations
- Lesson six (12th October 16:00-18:00). Coordination Strategies for Persistent Monitoring
- Static coverage based on Voronoi tessellations
- Control-based persistent monitoring
- Planning-based persistent monitoring
