Research And Implementation Of Teleoperation Vision Method Based On Augmented Reality

Teleoperation is the method that people operate machinery indirectly for some commission away from certain distance.Now,the mechanical intelligence cannot yet meet requirements,in some hazardous environment and commissions,such as nuclear industry,mining industry,explosion clearing,space exploration,etc.,remote operation of machine from a distance can ensure the life safety of operators,and give full play to the subjective judgement and operating experiences of people to deal with emergencies.Vision is main source for human to receive information of external condition,which is vital to enhance the operators' control of the far-end site situation.As a result,the research of visual telepresence for teleoperation is more and more extensive.However,in stereo vision space,compared with the same distance judgment in the real world,the selfcentered distance of perception is often underestimated,which means that there is the problem of depth perception being compressedIn this paper,the compress problem of depth perception existing in the teleoperation visual telepresence system is studied,and main work includes:(1)Design and complete the experimental system for the teleoperation visual telepresence.Stereo vision is provided for operator by the system,with the resolution of 1280x720,and average fps after adding the image processing algorithm is 22 frames,conforming to the real-time requirement.There is the function of head follow-up,and operator can control the shooting direction of remote binocular camera through the headturning locally.The max.Follow-up Speed of the pan-tilt is 360°/s and average error of 90° follow-up is 3.7%.The wireless transmitting of video function can make the far-end video picture be transmitted back to local through WIFI.Meanwhile,one set of remote controllable robot is equipped for teleoperation experiment.(2)Put forward the auxiliary information and operations,such as augmented reality combined with depth prompts and direction prompts.First,complete the image preprocessing such as standardization of binocular camera,stereo correction mapping table,etc.On this basis,calculate the depth of target by applying the method of target tracking and stereo matching based on characteristic,aiming at the problems that the problem that spatial perception is compressed and the distance between objects is shortened in visual telepresence;finally,the hybrid tracking registration method is used to get the attitude of the target,and the augmented reality method is used to integrate the depth and direction information into the stereo video.The experimental results show that the efficiency and accuracy of the operation are improved by the addition of augmented reality auxiliary information.(3)Applying the teleoperation vision system constructed in this paper,the accuracy of the depth calculated by this algorithm is tested,and the effectiveness of depth prompt method and direction prompt method is verified.The experimental results show that under the same conditions,adding depth information can effectively improve the operator's control of distance,while adding auxiliary information can improve the operation efficiency by 11.95%.