Simulation-based approach for avoiding external faults

Abstract

When a robot interacts with its environment to perform tasks, it often faces unexpected situations which render its actions unsuccessful despite perfect functioning of its components. These situations occur as deviations of properties of the objects (manipulated by the robot) from their acceptable values. Hence, they are experienced by the robot as external faults. In this work, we propose a simulation-based approach for avoiding the external faults that occur during the manipulation actions of a robot which involve releasing of objects. With the help of a single example simulation, that shows the behavior of the manipulated object for successful completion of the action, the proposed approach constructs different examples of the object's behavior and labels them as `desired' or `undesired'. These labelled examples are used by an algorithm, which we refer as N-Bins, to suggest a releasing state of the object that avoids the occurrence of external faults. Once exposed to the labelled examples, N-Bins can also be used for predicting the occurrence of external faults for a given releasing state of the object. These abilities of N-Bins are used by the approach to modify the releasing action of the robot for avoiding external faults. We present the approach as a four-step scheme that performs the above mentioned tasks completely autonomously.