Thèse / Université De Rennes 1 Ecole Doctorale Matisse Contributions to Control Modeling in Visual Servoing, Task Redundancy, and Joint Limits Avoidance. Acknowledgment

Visual servoing has become a popular paradigm in robot control by exploiting the informa-tion provided by a vision sensor in a feedback control loop. The research described in thisdissertation aims mainly to solve problems in visual servoing control and to improve theability to consider more effectively additional tasks. This thesis presents the state of art invisual servoing, redundancy, and joint limits avoidance and the following contributions: A new first order control scheme obtained by introducing a behavior parameter in a hy-brid control matrix is first proposed. It allows a better behavior of the system when suitablevalues for the behavior parameter are satisfied. An analytical analysis of the most commoncontrol scheme and the new one is performed when the camera displacement is a combina-tion of a translation along and a rotation around the camera optical axis. New second ordercontrol schemes are also proposed to enhance the behavior of the system when the problemof reaching a desired singular configuration is considered. Theoretical contributions regarding redundancy-based task-priority are achieved by develop-ing a new projection operator obtained by considering only the norm of the main task. Thisleads to a less constrained problem and enlarges the applicability domain of redundancy-based task-priority. New strategies for redundancy-based robot joint limits avoidance aredeveloped. The problem of adding additional secondary tasks to the main task while ensur-ing the joint limits avoidance is finally solved. All these works have been investigated experimentally using an eye-in-hand system andvisual servoing applications.