Research On Measurement And Control Of Omnidirectional Motion Platform Based On Computer Vision

An omnidirectional motion platform based on the principle of adaptive treadmills is an important part of virtual reality motion input devices.Efficient and non-contact acquisition of human walking information plays an important role in the control and adjustment of the omnidirectional motion platform.Research on computer vision-based measurement of human motion information has important theoretical and engineering practical value.Aiming at the problem that the spatial position of the user's body on the omnidirectional motion platform based on the adaptive treadmill principle does not change significantly from the ground.It is difficult to obtain the human body movement speed through the commonly used position difference method,a general,no need to mark A walking speed estimation method based on computer vision without wearing equipment.This method uses the spatial position data of human skeleton key points collected by the Kinect camera,and uses quaternion calibration correction,filtered noise Gaussian filtering and cubic spline interpolation technology to complete the missing data,and then obtains it with a step size correction algorithm.The gait spatiotemporal parameters of the human body when walking on the treadmill,and then the gait spatiotemporal parameters calculated above,to obtain the user's walking speed on the omnidirectional motion platform.The effectiveness of the proposed speed estimation algorithm is verified by comparing the speed estimation value with the speed setting value on a fixed-speed treadmill.The results show that the method can be used in subsequent studies related to human walking speed.Aiming at the application requirements of the combination of omnidirectional motion platform and virtual reality game,a method based on neural network to predict human walking posture is proposed.The error of the comparison between the predicted human body position and posture and the actual value can reach the centimeter level,and the data is transferred to the control system in advance due to the 0.5s transmission in advance,which effectively reduces the delay of the virtual reality system and increases the dynamic response speed of the system.Aiming at the problem that users may have unstable situations in virtual reality games,a method for calculating the human body support area and extrapolated centroid through human joint points is proposed to determine the stability criterion of the human body when walking on a treadmill.By adding multiple sinusoidal signal disturbances to the fixed-speed treadmill as interference,the unstable conditions of human walking are simulated,and the effectiveness of the proposed method is verified.The control strategy of the existing omnidirectional motion platform requires expensive sensor equipment or complex equipment transformation,and the acceleration and the omnidirectional motion platform structure occupy space that is difficult to balance.A control method based on Kinect camera and virtual spring damping is proposed.Analyze the control system through Simulink simulation.In addition,compare the control strategy adopted in this paper with the strategy adopted by the predecessors.The experimental results show that the control algorithm adopted in this paper has a faster response speed.