Generation of Virtual Tunnels for Meal Activity Perception Assist With an Upper Limb Power Assist Exoskeleton Robot

Various human assistance robots that effectively assist the daily activities of persons with perception deficits have recently been developed. In this paper, a perception assist method with an upper limb power assist exoskeleton robot is proposed for meal activities, which are important in daily living. The proposed robot-based method assists users by appropriately assessing utensil trajectories and posture performance according to the target position of the user. A method to predict the user’s target position based on the user’s motion trajectories and shoulder muscle electromyographic (EMG) signals is also proposed. Two kinds of virtual tunnels (for the utensil trajectory and utensil posture) corresponding to the user’s target position are defined in the proposed method to monitor the accuracy of the executed motion. The spatial relationship between the virtual tunnels is controlled to perform the eating task with the assistance of the power assist robot. The user’s hand motion is automatically modified by the robot in real time if the utensil trajectory and/or posture exits the respective virtual tunnel during meal activities. The effectiveness of the proposed method is evaluated through experiments.