Orion : Problem Solving Environments for autonomous systems.
Orion is a multi-disciplinary team at the frontier of computer vision, knowledge-based systems, and software engineering. The aim of Orion is to design and develop technics and softwares aimed at on the one hand automatic image interpretation and on the other hand program supervision and reuse. This objective is pursued by developing knowledge expression languages, as well as learning mechanisms and reasoning mechanisms for this knowledge, dedicated to specific classes of problems.
Orion focuses on the study of the knowledge involved in both classes of problems addressed : knowledge on objects and scenarios to recognize for automatic image interpretation and knowledge on the programs and on their use for automatic program supervision. We principaly study hybrid knowledge representation techniques (frame and rule based), temporal reasoning techniques and planning techniques. Symbolic learning is also a concern for knowledge base refinement.
The main application domain of Orion is computer vision. In a long term we intend to apply our methods and techniques to other domains (signal processing or scientific computation).