Automatic Design of Robot Behaviors through Constraint Network Acquisition

Control architectures, such as the LAAS architecture [1], CLARATY [12] and HARPIC [9], have been developped to provide autonomy to robots. To achieve a robot’s task, these control architectures plan sequences of sensorimotor behaviors. Currently carried out by roboticians, the design of sensorimotor behaviors is a truly complex task that can require many hours of hard work and intensive computations. In this paper, we propose a Constraint Programming-based framework to interact with roboticians during the sensorimotor behaviors design. A constraint network acquisition platform and a CSP-Based planner are used to automatically design sensorimotor behaviors. Moreover, our architecture exploits the propagation properties of the acquired CSPs to supervise the execution of a given sensorimotor behavior. Some experimental results are presented to validate our approach.

• Type: Conference
• Year: 2008
• Where: ICTAI
• Authors: Mathias Paulin, Christian Bessiere, Jean Sallantin
• Keywords: Artificial Intelligence, Control Architectures, ICTAI 2008, Sensorimotor Behavior, Sensorimotor Behaviors Design