Mobile robot obstacle avoidance method based on neural network

Abstract

The invention relates to a mobile robot obstacle avoidance method based on a neural network. The mobile robot obstacle avoidance method comprises the following steps of: (1) determining obstacle avoidance parameters according to the size and driving mode of a robot; (2) inputting a depth image, preprocessing the depth image, and segmenting a foreground region; (3) constructing an end-to-end obstacle avoidance neural network; (4) constructing a data set, and training an obstacle avoidance neural network; (5) acquiring a depth image, and carrying out the same preprocessing in the step (2) on thedepth image to obtain a foreground region; (6) performing large obstacle avoidance processing if a large obstacle exists in the foreground region and then executing a step (8), otherwise, executing astep (7); (7) inputting an image of the foreground region into the obstacle avoidance neural network, and outputting a turning angle and a moving speed of the robot; (8) and completing obstacle avoidance. According to the mobile robot obstacle avoidance method, the mobile robot obstacle avoidance is carried out by using the depth image and the convolutional neural network, manual feature extraction and parameter setting are not needed, and accurate obstacle avoidance can further be carried out in an outdoor complex scene.

Description

technical field [0001] The invention belongs to the field of robots, and in particular relates to an obstacle avoidance method for a mobile robot based on a neural network. Background technique [0002] Robots can replace humans in heavy and tedious physical labor, involving multidisciplinary knowledge such as machinery, electronics, sensors, computers, and artificial intelligence. There are various types and forms of robots. Currently, the mainstream forms on the market are mobile robots and arm robots. Among them, mobile robots have been active in factories, shopping malls and warehouses, and sometimes can also be used as mobile platforms for other robots, such as wheeled robots and mechanical arms to assist in tasks such as opening doors and serving tea and water, as well as wheeled robots and multi-purpose robots. The axis gimbal completes tasks such as target tracking and inspection. In the above two typical robot applications, the assisting work of the manipulator, t...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the collision problem that may occur when the robot is moving, the present invention provides a neural network-based obstacle avoidance method for mobile robots, which specifically includes the following six parts: robot obstacle avoidance parameter determination, input image preprocessing, end-to-end Neural network design for obstacle avoidance, data set production and network model training, obstacle avoidance for large obstacles, and neural network model reasoning for obstacle avoidance. The specific methods are as follows:

Images