Skip to Main content Skip to Navigation
Theses

Commande en suivi de chemin et en roulis des robots mobiles rapides en présence de glissements et d'instabilités

Abstract : Intervention robotics must meet the continuing need to go farther and faster. Within this framework, all-terrain mobile robots undergoing high velocities endure complex dynamic excitation including vibration, shock, impact and the resulting deterioration in quality of forces transmission in the wheel-ground contact that can lead to a loss of stability and hence undermine their mission. This thesis contributes to the development of control laws that ensure a robust path tracking besides a certain degree of stability (minimize the risk of rollover). In addition, the thesis proposes a new mechatronic device for active control of vehicle roll and increase the performance of mobility, and in particular its dynamic control during cornering. The control laws are developed based on physical models that take into account the vehicle dynamics and slippage phenomenon resulting from the wheel-terrain interaction. Validation and identification of these models are achieved from experimental results performed on the robot FAST-A (one of the two prototypes of the ANR-FAST project). These analytical models and experimental results are also compared to a complete 16 DOF numerical model, developed under the multi-body system environment MSC.ADAMS, in which the mechanisms of suspension, steering and traction, and the conditions of wheel-terrain contact are more accurately represented. The first control algorithm based on the LQR approach is used for the path-tracking problem. It considers a linear vehicle dynamics, and a linearized model of the kinematics of the robot in relation to its reference trajectory. The control algorithm is based on an optimization of tracking errors and system inputs. The performance of this control law is evaluated using the results from both the numerical simulation on MSC.ADAMS and the actual experience performed using the prototype. Furthermore, a comparison with another enhanced control law based on Model Predictive Control (MPC) techniques has been provided. This control technique has the ability to predict and anticipate future changes on the reference. The latter approach based on nonlinear continuous-time generalized predictive (NCGPC) proved to be more accurate in terms of path tracking and smother in terms of control signal. The applied method is also extended to any MIMO system with arbitrary numbers of inputs and outputs. Finally, a linear MPC is developed for the anti-roll device. This is based on an optimization on a finite time horizon of a certain criterion consists of a stability index and an index of energy consumption of the integrated device. The behaviour obtained is similar to the behaviour of a motorcycle rider who leans into the turn to counteract the centrifugal force. The lateral load transfer between the right and left wheel sides defines the stability index. Simulation results of this device are very promising with a remarkable reduction of the charge transfer of about 30%, thus increasing the safety of the device and finally a larger speed range (at a constant radius of curvature). The device in question is being assembled on the platform FAST-B (second prototype of ANR-FAST project) and will be tested during March and April 2012
Document type :
Theses
Complete list of metadata

Cited literature [59 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00831602
Contributor : Theses Bupmc <>
Submitted on : Friday, June 7, 2013 - 12:47:25 PM
Last modification on : Wednesday, December 9, 2020 - 3:09:17 PM
Long-term archiving on: : Sunday, September 8, 2013 - 4:18:46 AM

Identifiers

  • HAL Id : tel-00831602, version 1

Citation

Mohamed Larbi Krid. Commande en suivi de chemin et en roulis des robots mobiles rapides en présence de glissements et d'instabilités. Automatique / Robotique. Université Pierre et Marie Curie - Paris VI, 2012. Français. ⟨NNT : 2012PA066094⟩. ⟨tel-00831602⟩

Share

Metrics

Record views

566

Files downloads

3351