Skip to Main content Skip to Navigation

Dynamic Control and Singularities of Rigid Bearing-Based Formations of Quadrotors

Abstract : Bearing formation control allows groups of quadrotors to manoeuver in a desired geometry, using only visual measurements extractable from embedded monocular cameras. Prior works have treated quadrotors as single or double integrators, and as a result must operate slowly to compensate for unmodelled non-linearities. This thesis allows for faster bearing formations by developping higher-order controllers, considering the non-linear quadrotor and visual feature dynamics. A dynamic feedback controller based on second-order visual servoing and a model predictive controller are developped and tested in simulation and experiments, showing improved dynamic manoeuvering performance. The later is augmented with constraints such as field of view limitations and obstacle avoidance. All bearing formation algorithms depend on a sufficient degree of bearing rigidity to guarantee performance. This may be evaluated numerically, but as the rigidity is a function of the formation embedding, previous work could not guarantee rigidity in formations larger than a few robots. The second main contribution of this thesis is the evaluation of bearing rigidity singularities (i.e. embeddings where an otherwise rigid formation becomes flexible) by applying existing geometric analysis methods on an kinematic mechanism which is analoguous to the kinematic constraints imposed by the formation controller and robot models. This is extended to a novel classification system based on a contraction of constraint sets that can determine singular geometries for large formations, allowing for a formulation of a set of guaranteed rigid configurations without an ad-hoc kinematic analysis of individual formations.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Monday, June 20, 2022 - 12:50:09 PM
Last modification on : Friday, August 5, 2022 - 2:54:51 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03699595, version 1


Julian Erskine. Dynamic Control and Singularities of Rigid Bearing-Based Formations of Quadrotors. Automatic Control Engineering. École centrale de Nantes, 2021. English. ⟨NNT : 2021ECDN0044⟩. ⟨tel-03699595⟩



Record views


Files downloads