Patent
Obstacle detection method, device, apparatus and storage medium
Reads0
Chats0
TLDR
In this paper, an obstacle detection method, a device, an apparatus and a storage medium is presented. But the method comprises steps of acquiring acquired acquired 3D point cloud in a running process of an unmanned vehicle, projecting the 3d point cloud to two-dimensional grids and acquiring feature information of each grid; inputting the feature information from each grid into a prediction model obtained through the pre-training and obtaining an obstacle prediction parameter of the grid; and according to the obstacle prediction parameters, carrying out grid clustering so as to obtain the obstacle detection result.Abstract:
The invention discloses an obstacle detection method, a device, an apparatus and a storage medium. The method comprises steps of acquiring acquired 3D point cloud in a running process of an unmanned vehicle; projecting the 3D point cloud to two-dimensional grids and acquiring feature information of each grid; inputting the feature information of each grid into a prediction model obtained through the pre-training and obtaining an obstacle prediction parameter of each grid; and according to the obstacle prediction parameter of each grid, carrying out grid clustering so as to obtain an obstacle detection result. According to the invention, accuracy of detection results can be improved.read more
Citations
More filters
Patent
Method and apparatus for processing lane line data, computer device and storage medium
TL;DR: In this paper, a pre-trained deep neural network model was used to extract lane line regions and lane line attributes such as the dotted or full lane lines from the 3D point cloud data.
Patent
Method and device for switching driving modes
TL;DR: In this article, a deep learning decision model for switching between an end-to-end driving mode and a tracking driving mode by using safe scenario data and unsafe scenario data acquired by an onboard sensor of an automatic driving vehicle is presented.
Patent
Dynamic obstacle detection method, dynamic obstacle detection device, and computer readable storage medium
Liu Xin,Chen Xiao,Shan Dan +2 more
TL;DR: In this paper, a dynamic obstacle detection method consisting of acquiring the current vehicle speed of a current vehicle and the first position information of the current obstacle was proposed. But the method was not suitable for the high complexity and high calculation amount of a traditional dynamic obstacle detector.
Patent
Ground mark extraction method, model training method, device and storage medium
TL;DR: In this article, a mark-extraction network model is used to extract ground marks in a road segment map, and the ground marks are determined according to the ground mark information.
Patent
Three-dimensional point cloud tracking apparatus and method by recurrent neural network
Wang Chih-Wei,Tsai Ywe-Ting +1 more
TL;DR: In this paper, a three-dimensional point cloud tracking method using a recurrent neural network (RNN) was proposed to reconstruct the 3D point cloud of the entire environment at the current moment.
References
More filters
Patent
Simultaneous localization and mapping for a mobile robot
TL;DR: In this paper, a method of simultaneous localization and mapping includes initializing a robot pose and a particle model of a particle filter, each having an associated map, robot pose, and weight.
Patent
Traffic information extraction method based on laser sensor
TL;DR: In this article, a traffic information extraction method based on a laser sensor is proposed to provide accurate and effective road environment information in time, providing powerful safety control for driving of unmanned vehicles, and can also be used as a basis for making an unmanned vehicle driving strategy.
Patent
Road barrier identification method and device
TL;DR: In this paper, a road barrier identification method and device is presented, which comprises the steps of: obtaining laser point cloud of a barrier object on a road, obtaining mark information of the barrier object; and based on the mark information, generating training samples of a machine learning model.
Patent
Driverless vehicle, driverless vehicle positioning method, driverless vehicle positioning device and driverless vehicle positioning system
TL;DR: In this paper, the first laser point cloud height value data was matched with a current position of the driverless vehicle, where the first coordinates of all laser points and height values corresponding to the laser points in the first LPD data were obtained.