Reflective Understanding for Dependable Robots

To achieve autonomy in open, unstructured, changing environments while pursuing a specific mission ordered by human principals, robots shall control their bodies, adapt to different operational conditions, face unexpected situations, and decide about tradeoffs that may be presented to them. Achieving dependability implies keeping performance while guaranteeing other essential desiderata like safety, security or reduced environmental impact. In CORESENSE ( http://coresense.eu ) we seek a general, knowledge-based strategy to build resilient robotic systems. This strategy is based on the idea of making the robots understand their environments, themselves and their missions. We think that meagre understanding has always been the essential limitation of AI. In CORESENSE we target a science and technology of understanding to make robots more aware and capable of better dealing with the expected and the unexpected.