Research And Implementation Of Guided Robot Target Tracking And Motion Prediction

With the development of modern technology,service robots in public places have emerged in large numbers under the guidance of policies and market attraction.The service robots with the function of guiding guests have greatly influenced people’s lifestyle.Guidance robots need to perceive and judge the guests when performing guidance tasks.Therefore,this paper focuses on three aspects: target tracking,target positioning,and pedestrian motion prediction.This paper has carried on the target tracking algorithm.An improved kernel correlation filter(OURS)algorithm based on model and scale high-confidence update strategy was obtained.Using the method of one-dimensional fast discriminative scale space to solve the problem that the kernel correlation filtering algorithm can’t deal with the scale variation.A new confidence measurement method was proposed to judge the results of OURS algorithm model detection.When the target was completely occluded,it switched to Kalman filter algorithm to estimate the position of the target.In order to avoid the wrong scale update,it was proposed to judge the result of the scale detection with confidence.Experiments showed that the improved algorithm can effectively adapt to the scale variation,occlusion and deformation of the target and had better tracking performance than the original kernel correlation filter algorithm.This paper has carried on the positioning method.By comparing the filtering effects of median filter,mean filter,gaussian filter,bilateral filter and combined bilateral filter.The combined bilateral filter was selected to process the depth image,and the error was small.The depth probability histogram and pass-through filter were used to extract the information of the point cloud on the surface of the figure.Through the comparison of four kinds of target pedestrian depth direction positioning methods,more accurate depth direction coordinates were obtained.This paper has carried on the pedestrian prediction method.In view of the problem that the positioning accuracy of the Global coordinate system X-axis decreases obviously when the deformation of the target occurs,a polynomial fitting Kalman filter algorithm was proposed in this paper.Based on the constant acceleration model,the motion state model and observation model of the pedestrian served by the guidance robot were established,and the motion state system matrix,noise matrix and state transition matrix were determined.The pedestrian motion predicted by Kalman filter was realized.To achieve the pedestrian coordinate position,horizontal and longitudinal velocity of good prediction effect.The target tracking algorithm,target positioning algorithm and target motion prediction algorithm were implemented in the ROS robot operating system on the guidance robot platform,and nodes were set up for communication to realize the function of guiding the robot to guide the guests.The experimental results showed that the proposed algorithms were relatively stable in operation,and the proposed guidance framework was feasible.