Research On Collision-Free Safe Human-Robot Interaction System In Close Human-Robot Cooperation Scenario

Security issues have been an important constraint to the development of the human robot interaction field.Humans are able to sense the location of robots and avoid them at all times,while robots do not have this feature.This is the main reason why collisions occur during human robot interaction.To solve this problem,this paper proposes a collision-free safety human robot interaction system based on wearable inertial sensors.The system adds the function of sensing human position to the manipulator,and carries out the path planning of the manipulator so that the manipulator can avoid the human body while working,thus improving the safety of the human robot interaction process.The main research of this paper is as follows:(1)A wearable inertial sensor-based method for human upper body posture tracking.Current vision-based human perception methods are prone to occlusion problems and thus lose tracking of the human body.To address this problem,this paper proposes a real-time human upper body posture tracking method based on 5 inertial sensors.Firstly,the human body model is analyzed and a hinged human upper body skeleton model is established,then the inertial sensors are bound to the human body joints,and the inertial sensors are used to track the human body joint posture information,and the position information of the human upper body joint points is calculated through the human skeleton kinematic model.(2)Fast manipulator obstacle avoidance path planning method.In a collision-free safety human-robot interaction system,the manipulator needs to avoid the human body in real time,so the path planning time of the arm needs to be as short as possible.In this paper,several common RRT path planning algorithms are experimentally compared in a two-dimensional environment,and the shortest time-consuming RRT-connect algorithm is selected for manipulator path planning.Then the existing RRT-connect algorithm is improved for disadvantages such as large randomness and path redundancy.Finally,a fast manipulator planning method based on improved RRT-connect algorithm in joint space is realized.(3)Human and manipulator collision detection.The collision detection between human body and manipulator is the basis for the ability of the manipulator to evade the human body.In this paper,we obtain the human upper body posture information through inertial sensors in real time,and then convert the obtained human upper body posture information into an equivalent obstacle under the working space of the manipulator,and use the cylindrical envelope method to envelop the manipulator with the human body,which simplifies the collision problem between the human body and the manipulator into the collision detection of two cylinders in space.(4)Collision-free human-robot safety interaction system construction and visualization simulation software development.In this paper,the system is built in the context of a closerange human-robot collaborative assembly scenario.In order to have a better demonstration of the effect of the system,this paper develops the upper computer simulation software,including data communication and data processing parts,which can display the tracked human body and manipulator status in 3D real-time.Finally,the system is verified by simulation and physical experiments.Experiments show that the system can better track the human posture information,and the manipulator can recognize the human position and avoid during the movement.The safety and effectiveness of this collision-free safety human robot interaction system was verified.