Adaptive lidar point cloud correction and positioning method based on sweeping robot

Abstract

The invention provides an adaptive laser radar point cloud correction and positioning method based on a sweeping robot, comprising the following steps: S1. collect a new frame of data through the laser radar mounted on the robot; S2. collect through the wheel odometer Distance d and rotation angle θ, until the end of a cycle of collecting laser data frames, guess and estimate the pose of the robot corresponding to each odometer timestamp; S3. Considering the existence of errors, fit the pose of the robot, and calculate the fitting result After the mean square error; S4. Calculate the pose T of each laser point cloud relative to the sweeping robot k‑1 The coordinates P′ of k‑1 ; S5. Get the pose T of the sweeping robot k The coordinates of the laser point cloud above, and ICP matching of two adjacent frames of data, to obtain the pose T of the sweeping robot k . The invention adaptively selects a fitting model according to the mean square error result, and solves the problems of map distortion and inaccurate positioning caused by the distortion of the laser point cloud caused by the movement of the sweeping robot.

Description

technical field [0001] The invention relates to the field of robot positioning, and more specifically, to an adaptive lidar point cloud correction and positioning method based on a sweeping robot. Background technique [0002] With the development and progress of science and technology, the cost of lidar, which is often used in the military field, has been greatly reduced, which makes it possible for commercial applications. Lidar has good monochromaticity, high brightness, strong directionality, and anti-interference Strong, strong resolution, small and light equipment, etc., especially in speed measurement and robot positioning, it has a wide range of applications. The wheel odometer can quickly measure the moving distance and rotation angle, has high positioning accuracy in a short time and without slipping, and the encoder is cheap and suitable for commercial applications. As the first batch of commercial service robots, sweeping robots have entered thousands of househo...

AI Technical Summary

Problems solved by technology

[0003] The traditional sweeping robot only uses the lidar sensor for positioning. Since the robot will move or rotate during the scanning process of the lidar, the obtained laser point cloud and the generated map will have a certain degree of distortion, and the degree of distortion is the same as that of the robot. The speed of motion is directly proportional, especially the rotational motion has a great influence on the distortion of the laser point cloud, so the laser point cloud needs to be corrected; some odometer interpolation methods are used to correct the data, but only a few odometer data are used and ignored The overall motion of the robot (LOAM: Lidar Odometry and Mapping in Real-time) is not considered, and the existence of odometer measurement noise is not considered, so it cannot be corrected accurately

In engineering, rectilinear motion model correction is often used, but the correction effect on fast-moving robots is limited, and the robustness is not high

The polynomial fitting method is used to estimate the robot motion, but for a robot moving in a straight line, it often leads to overfitting, thus losing the overall information of the motion, and the correction results are inaccurate (Improved Techniques for Grid Mapping with Rao-BlackwellizedParticle Filters)

Images