Robot pose positioning method and system

Abstract

The invention discloses a robot pose positioning method and system. It relates to the field of robot positioning, wherein, the method collects IMU odometer data as a local reference system, obtains a previous moment pose state vector and a previous moment covariance matrix according to the local reference frame, and samples the last moment pose state vector, And perform unscented transformation on the sampling points, use the system model to predict the unscented transformed pose state vector and the covariance matrix at the previous moment to obtain the predicted value at the current moment, and combine the actual measured value to predict the predicted value at the current moment The filtering process obtains the relative pose measurement value at the current moment. After the filtering, the estimated value of the global pose at the current moment is obtained through coordinate transformation according to the measured value of the relative pose at the current moment, and the robot pose is positioned according to the estimated value of the global pose at the current moment. Using the unscented Kalman filter algorithm combined with the IMU odometer data and the actual measurement value collected by the GPS satellite or the vision system to obtain the global pose estimation value, which is robust to complex environments and improves the positioning accuracy.

Description

technical field [0001] The present invention relates to the field of robot positioning, in particular to a method and system for robot pose positioning. Background technique [0002] In the field of robotics, when it is necessary to solve the positioning problem in an unknown environment, the method of simultaneous localization and mapping (SLAM) is usually adopted. The core idea is to use sensor data to build a map of the surrounding environment in real time and realize the robot itself. position. The SLAM problem for robots is mainly the positioning problem of robots, which is essentially a state estimation problem. Therefore, the Kalman filter algorithm (KF) is usually used for state estimation to achieve positioning. However, the classical Kalman filter algorithm is only for linear systems, and practical problems are often based on more complex nonlinear systems. For nonlinear systems, the Extended Kalman Filter (EKF) algorithm, which is approximately linearized, is of...

AI Technical Summary

Problems solved by technology

However, the classic Kalman filter algorithm is only for linear systems, and practical problems are often based on more complex nonlinear systems.

For nonlinear systems, the Extended Kalman Filter (EKF) algorithm, which is approximately linearized, is mostly used. It uses the first-order Taylor expansion to deal with the linearization of nonlinear problems, ignoring the higher-order items of Taylor expansion, and there is a certain error in the solution.

Images