Indoor mobile robot visual positioning method based on convolutional neural network

Abstract

The invention discloses an indoor mobile robot visual positioning method based on a convolutional neural network, and belongs to the field of robot autonomous navigation. A foresight binocular camera and an airborne computer are carried on an indoor robot platform. The binocular camera collects images, feature points are extracted by using a convolutional neural network, and the feature points are used for a visual odometer based on a feature point method. And in the relocation, a target detection method is adopted to extract feature vectors of the pictures, and motion estimation is carried out. And the airborne computer constructs a local map for positioning by using the result of the visual odometer (or repositioning), and obtains the real-time pose of the robot. And the pose is fed back to a robot control system to control the position of the robot. According to the invention, the real-time pose estimation of the robot in the absence of GPS is realized, and the autonomous navigation level of the indoor robot is greatly improved.

Description

technical field [0001] The invention relates to automatic driving and positioning and navigation technology, in particular to a convolutional neural network-based visual positioning method for an indoor mobile robot. Background technique [0002] A mobile robot is a comprehensive system integrating environmental perception, dynamic decision-making and planning, behavior control and execution. It integrates multi-disciplinary research results in sensor technology, information processing, electronic engineering, computer engineering, automation control engineering, and artificial intelligence. It represents the highest achievement of mechatronics and is one of the most active fields of science and technology development. With the continuous improvement of robot performance, the application range of mobile robots has been greatly expanded, not only in industry, agriculture, medical care, service and other industries, but also in harmful and dangerous occasions such as urban sec...

AI Technical Summary

Problems solved by technology

The visual SLAM sensor uses a camera as a sensor, which is low in cost and simple in structure, and is suitable for large-scale promotion, but the biggest disadvantage is that it is greatly affected by the environment. With the construction of the map, there will be some cumulative errors and a large calculation load.

Images