Research On Human-robot Security Strategy Of Service Robot

With the rapid development of the service robot industry,the security of service robots has increasingly become the focus of attention.However,the current human-robot security strategies cannot satisfy the special working environment of service robots.Therefore,this thesis proposes a multi-level human-robot security strategy for service robots combined with vision and force awareness.The basic principle is to use the visual sensor to detect the shortest distance between the human body's joints and the robot's joints in real time.When it is less than the set safety threshold,the robot decelerates to a safe operating state and activates a force detection system.According to the information of force feedback,it is judged whether there is a collision between the human and the robot to execute the corresponding safety instructions.The safety between humans and robots can also be guaranteed without limiting the working space,quality,stiffness,and operating speed of the robot.Firstly,using a visual sensor to obtain a stable human skeleton model,this thesis uses a RGB-D camera to acquire the human body's original skeleton data.For the problem of human body occlusion,the position information of human joints is corrected according to the structural characteristics and kinematics of the human body.Finally,a stable human skeleton model is obtained by using a composite filtering algorithm.Secondly,the human-robot shortest distance detection method is studied.The kinematics analysis of the robot obtained the spatial position information of each joint of the robot,and the camera's calibration method is used to unify the spatial position information of the man-machine to a world coordinate system.The capsule-shaped cylinder bounding box model is established by using the spatial position information of humans and robots.Finally,the shortest distance between the human-robot joints is determined by calculating the positional relationship between the two bounding box models.Thirdly,the robot's collision recognition method based on force perception was studied,and the dynamic model of the robot was established using Newton-Euler method.The model can determine the theoretical moment of each joint when the robot is in motion;the actual torque of each joint is read through the joint torque sensors of the robot joints.Whether or not the robot has collided is determined based on whether the deviation between the two is within the error range.Then,based on the characteristics and advantages of vision and force security strategy,a multi-level human-robot security strategy of service robot based on visual-force perception is proposed,and the combination method of vision and force sense information is introduced.Finally,an experimental verification platform is built on the Baxter robot.The above-mentioned human-robot distance detection,robot collision detection and other algorithms are verified experimentally,and the results are analyzed to determine the validity of the proposed algorithm.