Real-Time Obstacle Avoidance Using Dual-Type Proximity Sensor for Safe Human-Robot Interaction

This letter introduces a dual-type proximity sensor and a control strategy for a robot manipulator to realize safe human-robot interactions (HRI) by using the sensor. Safety is an essential condition for HRI in practical scenarios. To achieve this condition, information about the relationship between an external objects and the robot is required. To obtain this information, we employ a dual-type proximity sensor, which consists of capacitive and inductive transducers and can detect the distance between a robot and external objects. Further, we propose a real-time trajectory planning method to deal with obstacles by using admittance control and distance measurements. To update the motion of the manipulator according to our control strategy, a Weight-Prioritized solution based on a QP (quadratic programming) formalism was applied. Further, the problem of self-sensing is solved via machine learning using a training dataset consisting of data corresponding to random joint positions. The proposed method was implemented on a collaborate robot (UR10). Experiments were conducted considering realistic human-robot interactions, and safety improvement was validated.

I. Introduction

Even a few years ago, manipulators were operated inside a fence and had to be stopped when workers approached them, and had to conform to ISO 10 218 [1]. Thus, there were numerous restrictions on the tasks that robots could perform and the places in which robots could be employed. However, collaborative robots, the so-called Cobot, which allow for complex tasks to be performed by humans and robots working together in the same space, have recently appeared. Accordingly, the demand for safe interaction technologies for human and robot collaborations is increasing in the field of robotics.