Robot life-long task learning from human demonstrations: a Bayesian approach

摘要: Programming a robot to act intelligently is challenging endeavor beyond the skill level of most people. Trained roboticists generally program robots for single purpose. Enabling be programmed by non-experts and perform multiple tasks are both open challenges in robotics. This paper presents framework that allows life-long task learning from demonstrations. To make possible, introduces representation based on influence diagrams, method transfer knowledge between similar tasks. A novel approach diagram presented along with demonstration teach an intuitive manner. The results three user studies validate enables simulated physical learn complex variety teachers, refining those during on-line performance, successfully completing different environments, transferring one another.