Image feasible region detection method, electronic device, storage medium, detection system

Abstract

The invention provides an image feasible region detection method, which includes obtaining a set of robot historical movement trajectories and a label identification, the label includes a feasible region and an obstacle region, and projecting all historical movement trajectory sets onto the collected image and marking them as a feasible region; laser data Projecting onto the collected image and marking it as an obstacle domain; training in the feasible domain, using the image and the label as a training sample, training to obtain a training model, and obtaining the corresponding label of a single pixel in the image according to the training model. The invention also relates to electronic equipment, storage media, and detection systems. The present invention builds a laser map under the global coordinates and fuses the trajectory information of the robot at different times, thereby expanding the feasible area and improving the navigation efficiency; and the process does not require human intervention, quickly generates a large number of samples, and is convenient for deployment in different environments.

Description

technical field [0001] The invention relates to the detection of the feasible region of a mobile robot, in particular to a method for detecting the feasible region of an image, an electronic device, a storage medium and a detection system. Background technique [0002] In recent years, with the continuous deepening of the research on the perception technology of outdoor mobile robots, the feasible region detection for robot navigation has become more and more important. The existing method is the feasible region detection method based on the supervised semantic segmentation neural network, which consumes a lot of manpower and material resources due to the need for a large amount of manually labeled data; especially the distribution of images in different environments is different, and the limited label data obtained in a certain place The trained neural network is difficult to expand to other places, which increases the difficulty of the actual deployment of the algorithm. ...

AI Technical Summary

Problems solved by technology

The existing method is the feasible region detection method based on the supervised semantic segmentation neural network, which consumes a lot of manpower and material resources due to the need for a large amount of manually labeled data; especially the distribution of images in different environments is different, and the limited label data obtained in a certain place The trained neural network is difficult to expand to other places, which increases the difficulty of the actual deployment of the algorithm

Images