Robot motion posture visual estimation method

Abstract

The invention discloses a robot motion posture visual estimation method. The method relates to the technical field of robot and machine vision, and comprises the steps of collecting continuous video images in a dynamic environment, selecting two of the collected images, detecting key points, calculating characteristic values of the key points, and detecting and storing the positions of the characteristic points; segmenting an original image, and obtaining and storing the position of a pixel region where a dynamic object is located in the image; comparing the image storing the feature point information with the image storing a segmentation result, and removing feature points distributed at the position of the pixel area where the dynamic object is located in the image storing the feature point information; and performing feature matching between the adjacent images by using the residual feature points after elimination and optimization, calculating pose motion of a robot between the adjacent images, and outputting robot motion pose estimation. The method has the advantages of being high in pose estimation accuracy, small in calculated amount and high in background environment interference resistance.

Description

technical field [0001] The invention relates to the technical field of robots and machine vision, in particular to a method for visually estimating robot motion postures. Background technique [0002] Mobile robots need to estimate their own motion posture in the working environment, which is especially important for mobile robots to achieve path planning, obstacle avoidance and other functions. Although the robot can obtain motion attitude estimation using high-precision inertial sensors, the cost is relatively high. At present, there are many interference factors in the estimation of robot motion attitude, and it is less involved in the estimation of robot motion attitude using only visual sensors. [0003] The present invention provides a method for visually estimating the motion posture of a robot, which greatly reduces the cost of sensor hardware and eliminates cumulative errors, and solves the problem that existing robot motion posture methods rarely use vision for es...

AI Technical Summary

Problems solved by technology

[0003] The present invention provides a visual estimation method for robot motion posture, which greatly reduces the cost of sensor hardware and eliminates cumulative errors, and solves the problem that the existing robot motion posture method seldom uses vision for estimation, and the robot simply uses visual sensors for motion posture Estimate the problem that is susceptible to interference, which has good practical value and engineering significance

Images