Object Grasping Assist System with Reading Human Intention and Predicting Object Position

In recent years, research on handing an object to a person using a robot arm has been increasing. However, the dynamic characteristics of the robot arm have been an issue in improving high-speed performance. In this study, we developed a system helps a person who has intention to grasp an object at high speed by bouncing the target object when it senses the human hand's intention to grasp it, especially for objects that are too far away to be reached by human hands. High-speed cameras are used to detect a hand and the target object motion on a desk. Next, when the hand begins to move, the solenoid actuator pushes to bounce the object in the direction of the human hand. At the same time, the laser pattern is projected on the desk to predict the destination of the target motion. Experimental results using a lightweight cup as the target object showed that a laser pattern was displayed around 450 ms ago before the moving object stops, and a series of movements were faster when the system was used than when the system was not used.