Robustness by Autonomous Competence Enhancement

The overall aim of this project is to develop an artificial cognitive system, embodied by a service robot, able to build a high-level understanding of the world it inhabits by storing and exploiting appropriate memories of its experiences. Experiences will be recorded internally at multiple levels: high-level descriptions in terms of goals, tasks and behaviours, connected to constituting subtasks, and finally to sensory and actuator skills at the lowest level. In this way, experiences provide a detailed account of how the robot has achieved past goals or how it has failed, and what sensory events have accompanied the activities.
Robot competence is obtained by abstracting and generalising from experiences, extending task planning and execution beyond preconceived situations. Activities successfully carried out by the robot for specific objects at specific locations may be generalised to activity concepts applicable to a larger variety of objects at variable locations. Conceptualisations may also result in commonsense insights, e.g. about object behaviour on tilted surfaces.
The project aims to produce the following key results:(i) \tRobots capable of storing experiences in their memory in terms of multi-level representations connecting actuator and sensory experiences with meaningful high-level structures,(ii)\tMethods for learning and generalising from experiences obtained from behaviour in realistically scaled real-world environments,(iii)\tRobots demonstrating superior robustness and effectiveness in new situations and unknown environments using experience-based planning and behaviour adaptation.
To achieve these ambitious goals, a consortium has been formed of research groups with long-standing expertise in high-level cognitive models, planning, learning, spatio-temporal knowledge representation, and robot sensing, navigation, and grasping. The consortium will establish a common conceptual framework for representing robot experiences, planning and learning. Results will be integrated and evaluated in an operational mobile platform with grasping facilities. We will demonstrate how a robot can evolve its understanding of the world as a result of novel experiences; and show how such understanding allows a robot to better cope with new situations and perform at a level of robustness and effectiveness not previously achievable.