Nouveaux concepts de locomotion pour véhicules tout-terrain robotisés

Robotic ground vehicles are mechanisms that use gravity and contact forces with the ground to perform motion. They can either be wheeled, tracked or legged. In this thesis we will focus on n-wheeled vehicles able to perform ground following motion with all the wheels maintaining contact at the same time. The main goal of this work is to establish the implication of the topological architecture of the vehicle mechanism on criteria such as climbing skills, robustness, ground clearance, weight, power consumption, and price. Efficient tools will be provided to help the robot designer to understand the implications of important design parameters like the number of wheels, the vehicle mechanism, and the motorisation of joints on the above criteria. The general state of a robotic ground vehicle can be described using spatial vectors containing both the linear and angular components of physical quantities such as position, velocity, acceleration and linear force. By definition, there is motion when the vehicle's link velocity state vector (expressed from the ground reference) is greater than zero. Wheeled ground following motion is then a special case of vehicle constrained motion where all wheels maintain contact with the ground. This thesis will describe a general kinematic and dynamic analysis of n-wheeled ground following robots. We will then discuss "contact forces optimisation techniques" and show the relationship between the number of wheels of a vehicle mechanism, the topological structure and the optimised degrees of fredom that we can get for the contact forces distribution. We will conclude with some considerations concerning the sensors needs for on-board terrain estimation. We will emphasise our argument using our two robot designs as examples: Shrimp: A 6-wheeled ground vehicle based on a 3 DOF passive suspension mechanism. With this design, no sensor based control is necessary to maintain ground contact with all the wheels. The distribution of tangential contact forces is done passively but can be optimised with on board active control and sensors for contact properties estimation (gyro, joint position sensors). Octopus: A 8-wheeled ground vehicle based on a (6 DOF active + 1 DOF passive) suspension mechanism. The autonomous coordination of the active 14 DOF is based on the on-board integration of inclinometer, joint position sensors and tactile wheels able to sense ground contact properties (angle, curvature, force, ...). With this design, active control can distribute the contact forces to minimise tangential forces and increase traction. This decreases the need for friction to climb obstacles. The theoretical investigation and new sensing concepts enable the design these two robots that demonstrate excellent capabilities for rough terrain. Passive Wheeled Locomotion Mechanisms (WLM) solutions are now mature enough for real applications like space exploration. However, active WLM solutions demonstrate potential climbing skills that cannot be equalled passively. Enhanced integration of sensors, actuators and advanced embedded control algorithms will lead to greater applications for future field and service robotics applications.