Skip to Main content Skip to Navigation
New interface

Contribution à la modélisation et à la commande de robots mobiles en présence de glissement : application au suivi de trajectoire pour les engins agricoles

Abstract : This dissertation addresses the topic of automatic guidance of mobile robots in natural en- vironment In particular path tracking on low adherent ground is investigated with a special focus on the experimental application automatic guidance of farm vehicles relying on a GPS sensor In such conditions the classical hypothesis of pure rolling without sliding is no longer relevant The numerous control laws built on such an hypothesis lead in this case to an important loss of precision unacceptable for the considered application The objective is thus to propose control algorithms making it possible to preserve the accuracy of the path tracking whatever the adherence conditions the configuration of the ground and the shape of the desired reference trajectory The design of relevant and tractable models is first described Because of the complexity in obtaining the parameters required in a dynamic model of vehicle (especially those linked to the tire/ground contact) dynamic models have been rejected for the control purpose Instead two alternative models called "extended kinematic models" have been proposed They allow to describe precisely vehicle performances - including sliding phenomena - with only two parameters to be estimated in real time Moreover such a point of view preserves a kinematic description suitable for the design of control algorithms In order to use these models the on-line estimation of these parameters is necessary It is achieved by two different kinds of estimator The first one relies on a direct calculation of the adherence properties (based on the difference between the actual process and the theoretical one when pure rolling conditions are respected) The second principle consists in observers design In order to apply this concept to the system considered the property of duality between control and observation is used to build robust observation algorithms with respect to the feature of the measurement system Based on these extended models two types of adaptive control scheme are built an Internal Model Adaptive Control as well as an Observer Based Control Both of them make it possible to control the vehicle with a satisfactory precision in steady operations (adherence conditions slowly varying) while preserving the advantages of the nonlinear control design developed for the case of rolling without sliding case In order to preserve the same high level of accuracy even during curvature transitions the delays induced by the low level actuators and the inertia of the vehicle have to be addressed (as they appear to be not inconsiderable in off-road vehicles) A model predictive control is so applied on the vehicle curvature servoing Using a model of the actuator it makes it possible to anticipate the variations of the reference path curvature and consequently to limit the overshoots (observed on tracking error) appearing at these transitions The overall control algorithms developped in this dissertation have been tested through actual experiments on an agricultural vehicle (farm tractor) in partnership with manufacturer CLAAS It appears that with the considered sensor (a GPS with a centimeter precision) the resulting tracking accuracy (included in a range of ± 1cm) is equivalent to the one obtained when the assumption of rolling without sliding is relevant This level of accuracy meets the expectations for driving assistance dedicated to agricultural applications.
Document type :
Complete list of metadata
Contributor : Camille Meyer Connect in order to contact the contributor
Submitted on : Thursday, March 29, 2012 - 5:08:46 PM
Last modification on : Tuesday, April 20, 2021 - 11:22:12 AM
Long-term archiving on: : Saturday, June 30, 2012 - 2:35:16 AM


  • HAL Id : tel-00683718, version 1


Roland Lenain. Contribution à la modélisation et à la commande de robots mobiles en présence de glissement : application au suivi de trajectoire pour les engins agricoles. Automatique / Robotique. Université Blaise Pascal - Clermont-Ferrand II, 2005. Français. ⟨NNT : 2005CLF21611⟩. ⟨tel-00683718⟩



Record views


Files downloads