In order to move easily in cluttered environments and/or achieve paths with small turning radius and with high agility, the mobile robot structure is designed such that the front and rear wheels can be steered and driven independently. This kind of four-wheel steering vehicle has proven to be interesting and promising for robotics applications, essentially for lateral dynamics control at high speed and for increasing the path curvature.
The planning and tracking problems have been divided into a two-level hierarchical controller. For the high-level controller, a new local path planning strategy is designed for obstacle-skirting in real time. This method computes rapidly and in real-time a smooth local path for obstacle avoidance that guarantees vehicle kinematic and dynamic constraints. The local path is automatically generated according to several waypoints based on the detected obstacle’s coordinates (here, we use a 3D LiDAR sensor for the perception of the environment surrounding the vehicle) and the current state of the robot (current position, steering angles, etc.). These waypoints are then connected by two cubic Bézier curves so that harsh curvatures and wide variation in steering angles are prevented.
This path is then fed to the low-level path follower synthesized by using a constrained Model Predictice Control (MPC) which is based on the vehicle dynamic model including sliding parameters. It is formulated as an optimization problem that computes at each time-step the optimal front and rear steering angles required to perform a desired path, with respect to multiple constraints, essentially the steering joint limits and the tire adhesion area bounds (i.e., pseudo-sliding zone limits where the grip conditions at wheel/ground are the best).
Moreover, This controllers includes wheel-ground lateral slippage and terrain geometry parameters that should be estimated online in order to enhance the path tracking performances. Thus, some observers are designed to estimate accurately these parameters in real-time using sensors measurements (e.g., RTK-GPS, IMU, etc.). All the proposed contributions have been validated through several tests on both advanced simulations under ROS/GAZEBO and experiments on a real off-road mobile robot at high and low velocities.
Watch the video, recorded during the scientific seminar run by Robagri in the 2019 FIRA. This project was presented by Mohamed FNADI, phD candidate at Sorbonne University.