Laser monocular vision fusion positioning mapping method in dynamic scene

Abstract

The invention discloses a laser monocular vision fusion positioning mapping method in a dynamic scene, and the method comprises the steps: 1, detecting dynamic obstacle information according to point cloud data of a current frame and pose priori information predicted by a visual odometer of a previous frame; 2, forming an image mask on monocular vision 2 with mutually calibrated external parameters, extracting ORB feature points on the image mask and matching the ORB feature points with ORB feature points of a previous frame, estimating the depth of the ORB feature points, obtaining a relative pose through pose calculation, and outputting key frame information meeting requirements; 3, inserting the key frame into the common view, updating the connection relation between the key frame and other key frames, the growth tree and the bag-of-word model according to the common view degree of map points, generating new map points, finding the adjacent key frame according to the essential map of the current key frame, constructing a nonlinear optimization problem, and optimizing the pose and the map points of the key frame; 4, judging whether the similarity between the image data of each key frame and the image data of the current key frame reaches a threshold value or not, if yes, judging that loopback occurs, replacing or filling map points where conflicts exist between the current key frame and the loopback key frame, then connecting the current key frame and the loopback key frame on the essential map, updating the essential map, and finally, carrying out global BA to obtain optimized key frame poses, feature point maps and point cloud maps. According to the invention, the SLAM purpose can be executed in a dynamic scene.

Description

technical field [0001] The invention relates to a laser monocular vision fusion positioning and mapping method for the field of automatic driving, in particular to a laser monocular vision fusion positioning and mapping method in a dynamic scene. Background technique [0002] With the rapid development of technologies such as automation, artificial intelligence and robotics, more and more intelligent unmanned systems with perception, positioning and navigation functions are applied in real life and industrial scenarios, such as sweeping robots, storage robots and driverless taxis car etc. With the continuous iteration of sensor technology and intelligent algorithms, the application scenarios of these unmanned systems have expanded from simple known environments to completely unknown complex environments. In the unknown environment, the robot needs to use the information from the sensor to perceive the surrounding environment and estimate its own state, so as to ensure that ...

AI Technical Summary

Problems solved by technology

Therefore, the robot will face three problems in an unknown environment: 1) where am I? 2) What is around me? 3) Where am I going? These three problems correspond to: 1) autonomous positioning of the robot; 2) map construction of the robot; 3) real-time obstacle avoidance and path planning of the robot

At present, most of the research on visual SLAM problems is based on the assumption that the surrounding environment is static, and the laser monocular vision fusion positioning and mapping method in dynamic scenes is still immature, and the relevant theories need to be further improved

Images