Human Grasp Classification for Reactive Human-to-Robot Handovers (2003.06000v1)

Abstract: Transfer of objects between humans and robots is a critical capability for collaborative robots. Although there has been a recent surge of interest in human-robot handovers, most prior research focus on robot-to-human handovers. Further, work on the equally critical human-to-robot handovers often assumes humans can place the object in the robot's gripper. In this paper, we propose an approach for human-to-robot handovers in which the robot meets the human halfway, by classifying the human's grasp of the object and quickly planning a trajectory accordingly to take the object from the human's hand according to their intent. To do this, we collect a human grasp dataset which covers typical ways of holding objects with various hand shapes and poses, and learn a deep model on this dataset to classify the hand grasps into one of these categories. We present a planning and execution approach that takes the object from the human hand according to the detected grasp and hand position, and replans as necessary when the handover is interrupted. Through a systematic evaluation, we demonstrate that our system results in more fluent handovers versus two baselines. We also present findings from a user study (N = 9) demonstrating the effectiveness and usability of our approach with naive users in different scenarios. More results and videos can be found at http://wyang.me/handovers.

• The paper presents a systematic evaluation method for handovers using freeform, guided, and distracted grasp strategies.
• It examines how variations in human grasp and distraction levels affect the robot’s responsiveness and efficiency.
• Results highlight that structured grasp demonstrations enhance the accuracy and adaptability of human-robot collaborations.

A System for Human-Robot Handover Task Evaluation

The paper presents a systematic evaluation process for testing human-robot handover interactions, focusing on a task involving the transfer of colored blocks from a human to a robot. This work emphasizes the importance of evaluating humanoid robotic systems in cooperative tasks, particularly in scenarios where robots must safely and efficiently accept items handed to them by human counterparts.

The experimental setup is designed to assess several parameters critical to successful human-robot interactions. The key experimental conditions tested include freeform handover, demonstration of predefined grasp strategies, and distracted task completion. These distinct phases enable a comparative analysis of human-robot handover efficiency under varying degrees of human attention and instruction.

1. Freeform Task Performance: Initially, participants are asked to transfer blocks to the robot using any grip or method of their choice. This stage is critical in understanding natural human tendencies and adaptations when interacting with robotic systems. The analysis of handover effectiveness and efficiency provides a baseline for subsequent evaluations.
2. Strategy Demonstration and Task Compliance: In a structured follow-up, participants are exposed to designated grasp approaches that the robot has been programmed to recognize. Demonstrations include grasp types such as open-palm hold and various pinch techniques. This phase examines how the inclusion of guided handover methods influences the fluidity and speed of the interaction, with implications for refining robotic programming to better accommodate human actions.
3. Distracted Interaction Assessment: The experiment further explores handover dynamics by introducing a distraction component. Participants are asked to watch a video and complete a cognitive task while simultaneously executing the handover, presenting a dual-task challenge that mirrors real-world scenarios. This stage investigates the robustness of the robot's detection and response mechanisms under conditions where human attention is divided.

The results of these experiments are anticipated to shed light on several practical considerations, such as the robot's responsiveness to different grip styles and human movement speed, as well as its capacity to interpret human intention when visual focus is shifted elsewhere. Such insights are invaluable for optimizing human-robot collaboration across diverse applications, from industrial settings to personal assistance roles.

The approach outlined demonstrates a comprehensive method for evaluating human-robot handover tasks, pushing towards designing robots that can seamlessly integrate into environments requiring close human-robot cooperation. Future developments may involve refining sensory input and cognitive algorithms within robotic systems to enhance their adaptability and predictive capabilities.

In summation, this paper underscores the significance of structured human-robot interaction studies, inviting further research into adaptable, context-aware robotic systems designed to work in synergy with human operators.