Attentive and corrective feedback for adapting robot's perception on fuzzy linguistic information

This paper proposes a method for understanding the fuzzy linguistic information based on the user's guidance. A quantitative assessment for a fuzzy linguistic term such as "little" depends on the environmental conditions. Therefore the corrective feedbacks are utilized to adapt the robot's perception toward the corresponding environment. However, the attentive commands like "move carefully" are used to modify the evaluation process of the fuzzy linguistic information according to the user's desire. The primitive behaviors are evaluated by a behavior evaluation network (BEN) and a feedback evaluation system (FES) is utilized to evaluate the corrective feedbacks. An attention level controller is introduced to change the attention level based on the attentive commands. The system is adapted toward the user's perception on the fuzzy linguistic information based on the corrective feedbacks in the adaptation phase. Then the attentive commands are used to modify the evaluation process according to the local requirements. The proposed system is demonstrated by using a PA-10 robot manipulator and a situation of cooperative rearrangement of user's working space is considered. The adaptation of the system toward the environmental conditions by the corrective feedbacks and the capability to use the attentive feedbacks to modify the system during the work, improve the effectiveness of the system.