University of Hamburg
Evaluation Metrics for an Experience-based Service Robot System
In a dynamic and changing world, a robust and effective robot system must have adaptive behaviors, incrementally learnable skills and a high-level conceptual understanding of the world it inhabits, as well as planning capabilities for autonomous operations. My talk will first demonstrate how an intelligent system like a robot can evolve its model as a result of learning from experiences; and how such a model allows a robot to better understand new situations by integration of knowledge, planning and learning in the framework of EU collaborative project RACE. Then I will show some integrated results of operational service robot platforms with grasping facilities in a restaurant service scenario. To evaluate success for a given task in a given scenario, we measure the compliance of the actual robot behavior to the intended ideal behavior for that task in that scenario by a measure of ‘‘Distance to Ideal Model’’. Additionally, we also measure the Description Length of the instructions given to the robot to achieve a goal matters. Our general aim of designing learning and reasoning tools for a robot to autonomously and effectively increase its competence was operationalized as: make it possible for a robot to collect experiences allowing it to perform at lower Distance to Ideal Model and shorter Description Length.