An obstacle detection method for unmanned vehicles based on tegrax1 radar data

Abstract

The invention discloses a driverless car obstacle detection method based on TegraX1 radar data, and the method comprises the steps: 1, employing a velodyne laser radar as a sensor for collecting the environment information, and carrying out the three-dimensional radar data conversion through an NVIDIA Tegra X1 mobile processor; 2, carrying out the obstacle detection based on grids, and employing a GPU to process grid data, wherein the step comprises three substeps: enabling three-dimensional data points to be projected to a grid map, enabling all grids with the relative heights being greater than one threshold value to be set as obstacle points, and filtering out all grids which are determined as the obstacles because there are suspension points in the grids. According to the technical scheme of the invention, the method employs the velodyne laser radar as the sensor for collecting the environment information, and carries out the optimization based on the NVIDIA Tegra X1 mobile processor through the GPU, and achieves the speeding up of the obstacle detection for a driverless car.

Description

technical field [0001] The invention belongs to the field of unmanned driving, and in particular relates to an unmanned vehicle obstacle detection method based on Tegra X1 radar data. Background technique [0002] In recent years, with the continuous development of sensor technology, control system, and artificial intelligence, ground mobile robots have made great progress. In a real dynamic environment, autonomous robots can stably and accurately detect obstacles and identify obstacle types in environmental perception, which can be of great help in establishing motion models for path planning, so as to make intelligent decision-making behaviors. Usually there are two main categories of dynamic objects in the environment of autonomous robots: vehicles and pedestrians. For the vehicle, it is the main interactive object in the traffic, the speed is relatively fast, and a certain safety distance needs to be maintained. When the traffic rules allow, it is necessary to choose wh...

AI Technical Summary

Problems solved by technology

However, because the radar data is too large, the processing algorithm has a huge amount of calculation, so the real-time performance is poor. Therefore, in a complex environment, the obstacle information cannot be sent to the decision-making system in time.

Images