Method and device for generating obstacle motion information for unmanned vehicles

Abstract

The embodiment of the present application discloses a method and a device for generating obstacle motion information for an unmanned vehicle. A specific implementation of the method includes: obtaining the current frame obstacle point cloud and the reference frame obstacle point cloud representing the target obstacle to generate motion information; calculating the target obstacle in each of the M first displacement observations The first observed displacement under the first displacement observation amount; determine the movement information of the target obstacle under each of the first displacement observation quantities in the M first displacement observation quantities; according to the determined M kinds of movement information and the target obstacle The historical motion information determines the observed motion information of the target obstacle; the current motion information of the target obstacle is generated by using the preset filtering algorithm with the motion information of the target obstacle as the state variable and the observed motion information as the observation quantity. This embodiment can realize effective motion estimation for obstacles in the case of inaccurate point cloud segmentation of obstacles.

Description

technical field [0001] The present application relates to the technical field of unmanned vehicles, in particular to the technical field of obstacle motion estimation, and in particular to a method and device for generating obstacle motion information for unmanned vehicles. Background technique [0002] Unmanned vehicles, also known as "mobile wheeled robots", conduct comprehensive analysis and processing based on the information collected by various sensors (such as cameras and laser radars) through the driving control equipment set on the vehicle to achieve Path planning and driving control. Most unmanned vehicles are equipped with lidar to collect external information. In the process of path planning and driving control for unmanned vehicles, obstacle detection can be performed on each frame of the laser point cloud collected by the lidar (that is, the laser point cloud collected by each sampling period of the lidar), and then , and then perform motion estimation on the...

AI Technical Summary

Problems solved by technology

[0003] However, most of the existing methods for obstacle motion estimation adopt the definition of interest points on the obstacle point cloud (point cloud used to characterize obstacles), and perform motion estimation on obstacles according to the displacement of interest points, thus There is a problem of inaccurate motion estimation when the obstacle point cloud segmentation is inaccurate (eg, under-segmentation or over-segmentation)

Images