Controlling dynamic contact transition for nonholonomic mobile manipulators

Vincent Padois (1) , Pascale Chiron (1) , Jean-Yves Fourquet (1)

1 LGP - Laboratoire Génie de Production

This work is devoted to planning and execution of complex missions in robotics. Robotics has evolved from an industrial, repetitive framework to application domains with much more variability of tasks, with increasing complexity in uncertain environment. This is clearly the case for service robotics for example, but even industrial robots have now to work in environment not totally calibrated for the task they have to perform. The result is that the classical decomposition in static ordered local tasks cannot apply in presence of such a variability. It has a poor dynamic performance and cannot cope with uncertainties. Our work is organized around a complex mission: "Go to the blackboard and write" for mobile manipulators that have capabilities of locomotion and manipulation. It is a simple and intuitive example of a complex mission that relies on different sensors, exhibits different operating modes and needs to switch between different feedbacks and set-points. Our approach is based on hybrid dynamical systems. It is focused on dynamical sequencing of control laws that ensures good transients, robustness and allows to update the mission at every transition from one mode to another. Simulations have been realized with Matlab Simulink and Stateflow toolboxes and an experimental validation has been developed within the controller on the H2bis nonholonomic mobile manipulator.

Domaines: Robotique [cs.RO]

Vincent Padois, Pascale Chiron, Jean-Yves Fourquet. Controlling dynamic contact transition for nonholonomic mobile manipulators. IEEE/RSJ International Conference on Intelligent Robots and Systems, Sep 2004, Sendai, Japan. pp.3817 -- 3822 vol.4, ⟨10.1109/IROS.2004.1390009⟩. ⟨hal-00624398⟩