Loading...

Fault Detection in Robotic Swarm

Mollahossein, Mojtaba | 2019

518 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52342 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Alasti, Aria; Salarieh, Hassan
  7. Abstract:
  8. Fault detection in swarm robots is one of the most important issues that has drawn a lot of scientists’ attention. Previous studies have shown a lot of problems regarding this issue, such as dependency of data-based methods to mission and high complexity of model-based methods on the condition that nonlinear models are used. Extended information filter is an applicable method for fault detection because it could turn the inverse of a big matrix to a simple summation of information that is so desired in terms of calculations. The goal of this study is to identify the agents with fault in swarm robots online by modifying the extended information filter. In this study, it has been attempted to develop a fairly comprehensive distributed and model-based (i.e. nonlinear model in extended information filter) method for fault isolation in sensors of swarm robots by combination of absolute and relative sensors. The proposed method is not dependent on mission and is scalable so that it could be used in any group of robots with any number of members. In this study, shape formation in robots has been considered as mission. First, in this mission, problems caused by faults is examined. Second, the extended information filter is introduced as an applicable method of fault detection and then, its problems are mentioned. Then, the problems are addressed and the project is explained in detail. To this end, absolute position sensors are increased, which leads to error reduction in estimation of absolute states, turning the central algoritm into the distributed algorithm, and performing calculations in all robots' processors. Then, a logic for fault isolation is introduced. Finally, the performance of the proposed algorithm is verified by a simulation of seven robots in shape formation. The results show that the proposed algorithm could isolate all of the faults in robots' sensors
  9. Keywords:
  10. Swarm Robot Formation ; Fault Detection ; Control ; Model Based Control ; Fault Isolation

 Digital Object List

 Bookmark

No TOC