Skip to Main content Skip to Navigation

Recongurable cooperative control for extremum seeking

Abstract : This thesis addresses the localisation of the maximum of an unknown spatial field in a delimited area. The search is performed by a multi-agent system composed of autonomous vehicles. The mission can be divided in two parts, the first one focuses on the estimation methods for optimisation, and the second one concerns the control law to move the fleet of agents. Two solutions have been proposed for the estimation part. The first one relies on a local search strategy that estimates the gradient of the unknown field and moves the agents along the gradient direction. The optimal sensor placement of the agents has been investigated and three criteria have been proposed to find the formation shape required for efficient estimation. Moreover, a sensor fault detection and isolation scheme using an adaptive threshold has been presented. The second estimation solution is a global search strategy based on a Kriging model of the field. A new sampling criterion is defined for the multi-agent system to locate the position of the global maximum while limiting the number of measurements and taking into account the agent dynamics. Both solutions provide a set of desired sampling positions to the agents. A distributed control law has been designed to guide the agents toward these locations. This control law is also used in the local approach to gather the agents in a desired formation and reconfigure it when a fault has been detected, following the optimal sensor placement analysis. The same control law has been adapted to reach the positions specified iteratively by the Kriging-based global search strategy.
Document type :
Complete list of metadata

Cited literature [120 references]  Display  Hide  Download
Contributor : Cécile André Connect in order to contact the contributor
Submitted on : Friday, February 26, 2016 - 3:03:12 PM
Last modification on : Thursday, October 28, 2021 - 3:16:37 PM
Long-term archiving on: : Sunday, November 13, 2016 - 3:50:19 AM


  • HAL Id : tel-01279560, version 1



Arthur Kahn. Recongurable cooperative control for extremum seeking. Automatic. UNIVERSITE PARIS-SACLAY, 2015. English. ⟨tel-01279560⟩



Les métriques sont temporairement indisponibles