Exploring the interaction strategy and release timing for robot-to-human handovers with manually guided motion

Object handover is one of the fundamental tasks of service robots. This paper focuses on a robot-to-human object handover controller applied to a domestic service robot. People in need often have manual operation constraints caused by different body postures or declined physical functions. The preplanned handover strategy used in previous studies potentially increases their cognitive burden and makes it difficult to meet their needs for flexible real-time control and intuitive interaction. It also remains challenging to deal with sensing and prediction errors, motion planning and coordination. The robot no longer provides assistance after an object is delivered to a preplanned handover location, and the receiver needs to make extra efforts to get the object released. Therefore, inspired by a human study, a handover controller is designed based on a manually guided handover strategy that solve problems from grasp adjustment to object release in dynamic retraction motion. Flexible control and intuitive interaction are enabled through motion with continuous support from the robot. Without additional sensors or modeling learning, changes in motion and energy consumption are taken into account to explore an appropriate release timing. The results of robot-to-human handover experiments and the user studies indicate that the receiver can freely pull an object to a suitable location as desired and smoothly obtain the object’s control to complete a subsequent task using the proposed controller with proper release timing.