Abstract: The present disclosure discloses a method, apparatus and server for constructing a map. An embodiment of the method comprises: building a point cloud data frame sequence using point cloud data frames acquired in real time; extracting morphological data from the point cloud data frames; establishing a spatial position relation between the morphological data of two adjacent point cloud data frames in the point cloud data frame sequence; determining a reference spatial position relation corresponding to a stationary object based on the spatial position relation, and constructing a map with the reference spatial position relation. The present embodiment realizes constructing a map with the point cloud data.

Abstract: Disclosed embodiments include a method and a device for generating position information of a target object. In some embodiments, the method comprises: obtaining target point cloud data of the target object, acquired by a target scanner at a target position, and position information of the target scanner, and obtaining reference point cloud data of the target object, acquired by a reference scanner at a reference position, and position information and an Euler angle of the reference scanner; obtaining an Euler angle of the target scanner based on the target point cloud data, the reference point cloud data, and the Euler angle of the reference scanner; and generating the position information of the target object based on the target point cloud data, the position information and the Euler angle of the target scanner, the reference point cloud data, and the position information and the Euler angle of the reference scanner.

Abstract: The present disclosure discloses a method and apparatus for identifying laser point cloud data of an autonomous vehicle provided with a lidar. A specific implementation of the method comprises: acquiring current pose information of the autonomous vehicle in a world coordinate system in response to receiving a latest frame of laser point cloud data collected by the lidar; acquiring from the cache, based on the current pose information, N×N map blocks centered on map blocks corresponding to the current pose information in a preset three-dimensional grid map and pre-loaded into a cache and are; and executing, for each laser point data in the received laser point cloud data, the laser point data identification operations. The implementation realizes identifying whether each laser point data is a static laser point, and can improve the accuracy rate of identifying a laser point data obstacle.

Abstract: Disclosed embodiments include an unmanned vehicle, a method, apparatus and system for positioning an unmanned vehicle. In some embodiments, the method includes: acquiring first laser point cloud height value data matching a current position of the unmanned vehicle, the first laser point cloud height value data; converting the first laser point cloud height value data into laser point cloud projection data in a horizontal earth plane; determining a first matching probability of the laser point cloud projection data in a predetermined range of a laser point cloud height value map; and determining a position of the unmanned vehicle in the laser point cloud height value map based on the first matching probability. The embodiment implements an accurate positioning on the current position of the unmanned vehicle.

Abstract: The present disclosure discloses a method and apparatus for acquiring information. A specific embodiment of the method includes: collecting characteristic information of a road object, the characteristic information comprising: a position, in a world coordinate system, of laser points corresponding to the road object in a laser point cloud collected in an area where the road object is located, and a position in a reflectance map corresponding to a position of a control point of the road object in the world coordinate system; and determining whether a preset accuracy is acceptable based on a match between the characteristic information of the road object and a preset characteristic information of the road object, the preset accuracy comprising: a splicing accuracy of the spliced laser point cloud, an accuracy of the reflectance map, and an accuracy of the high-precision map.

Abstract: The present application discloses a method and apparatus for positioning a vehicle. An implementation of the method comprises: obtaining laser point cloud data of a laser point cloud collected by a laser radar on a vehicle, and obtaining an initial pose of a center point of the laser radar; calculating a matching probability between projection data corresponding to each sampling pose and map data of a reflected value map respectively; and calculating an optimal pose based on the matching probability between the projection data corresponding to the each sampling pose and the map data of the reflected value map, and determining a position of the vehicle based on the optimal pose.

Abstract: The present application discloses a method and apparatus for updating a map. An implementation of the method comprises: acquiring a laser point clouds obtained by scanning a surrounding environment and acquiring a periodic pose change of a vehicle in a scanning period; determining an a priori pose of the laser radar at a reference time in the current scanning period based on a known pose of the laser radar in a previous scanning period and the periodic pose change; performing motion compensation on the scanned laser point clouds to form a compensated point cloud frame; generating a partial point cloud characteristic map; matching the partial point cloud characteristic map with a map area obtained in a reference point cloud map, and correcting the a priori pose to obtain an a posteriori pose; and updating a three-dimensional point cloud map of a to-be-constructed area using the compensated point cloud frame.

Abstract: The present disclosure discloses a method and apparatus for positioning a vehicle. In some embodiments, the method comprises: acquiring an a priori position of a to-be-positioned vehicle at a current positioning moment determined by performing a strapdown calculation between a previous positioning moment and the current positioning moment; determining a map area for searching in a laser point cloud reflected value map; matching a reflected value characteristic of a projection area generated by projecting a real-time laser point cloud, to obtain a map matching position according to a matching result; positioning, using the a priori position in combination with observation data of a vehicle-mounted global navigation satellite system (GNSS) receiver of the vehicle, to obtain a satellite positioning position; and fusing the a priori position, the map matching position and the satellite positioning position to generate a positioning result of positioning the vehicle at the current moment.

Abstract: Disclosed embodiments include a driverless vehicle, and a method, an apparatus and a system for positioning a driverless vehicle. In some embodiments, the method includes: acquiring first laser point cloud reflection value data matching a current position of the driverless vehicle; converting the first laser point cloud reflection value data into laser point cloud projection data in a horizontal earth plane; determining a first matching probability of the laser point cloud projection data in a predetermined range of a laser point cloud reflection value map by using a position of a predetermined prior positioning position in the laser point cloud reflection value map as an initial position; and determining a position of the driverless vehicle in the laser point cloud reflection value map based on the first matching probability. The implementation can accurately position the current position of the driverless vehicle.

Abstract: Systems and methods are described for multithreaded navigation assistance by acquired with a single camera on-board a vehicle, using 2D-3D correspondences for continuous pose estimation, and combining the pose estimation with 2D-2D epipolar search to replenish 3D points.

Abstract: Disclosed embodiments include an unmanned vehicle, a method, apparatus and system for positioning an unmanned vehicle. In some embodiments, the method includes: acquiring first laser point cloud height value data matching a current position of the unmanned vehicle, the first laser point cloud height value data; converting the first laser point cloud height value data into laser point cloud projection data in a horizontal earth plane; determining a first matching probability of the laser point cloud projection data in a predetermined range of a laser point cloud height value map; and determining a position of the unmanned vehicle in the laser point cloud height value map based on the first matching probability. The embodiment implements an accurate positioning on the current position of the unmanned vehicle.

Abstract: Disclosed embodiments include a driverless vehicle, and a method, an apparatus and a system for positioning a driverless vehicle. In some embodiments, the method includes: acquiring first laser point cloud reflection value data matching a current position of the driverless vehicle; converting the first laser point cloud reflection value data into laser point cloud projection data in a horizontal earth plane; determining a first matching probability of the laser point cloud projection data in a predetermined range of a laser point cloud reflection value map by using a position of a predetermined prior positioning position in the laser point cloud reflection value map as an initial position; and determining a position of the driverless vehicle in the laser point cloud reflection value map based on the first matching probability. The implementation can accurately position the current position of the driverless vehicle.

Abstract: Methods and systems for moving object localization include estimating a ground plane in a video frame based on a detected object within the video frame and monocular structure-from-motion (SFM) information; computing object pose for objects in the frame based on the SFM information using dense feature tracking; and determining a three-dimensional location for the detected object based on the estimated ground plane and the computed object pose.

Abstract: A method for performing three-dimensional (3D) localization requiring only a single camera including capturing images from only one camera; generating a cue combination from sparse features, dense stereo and object bounding boxes; correcting for scale in monocular structure from motion (SFM) using the cue combination for estimating a ground plane; and performing localization by combining SFM, ground plane and object bounding boxes to produce a 3D object localization.

Abstract: Systems and methods are disclosed for multithreaded visual odometry by acquired with a single camera on-board a vehicle; using 2D-3D correspondences for continuous pose estimation; and combining the pose estimation with 2D-2D epipolar search to replenish 3D points.

Abstract: Methods and systems for moving object localization include estimating a ground plane in a video frame based on a detected object within the video frame and monocular structure-from-motion (SFM) information; computing object pose for objects in the frame based on the SFM information using dense feature tracking; and determining a three-dimensional location for the detected object based on the estimated ground plane and the computed object pose.

Abstract: Systems and methods are disclosed for autonomous driving with only a single camera by moving object localization in 3D with a real-time framework that harnesses object detection and monocular structure from motion (SFM) through the ground plane estimation; tracking feature points on moving cars a real-time framework to and use the feature points for 3D orientation estimation; and correcting scale drift with ground plane estimation that combines cues from sparse features and dense stereo visual data.

Abstract: A method for performing three-dimensional (3D) localization requiring only a single camera including capturing images from only one camera; generating a cue combination from sparse features, dense stereo and object bounding boxes; correcting for scale in monocular structure from motion (SFM) using the cue combination for estimating a ground plane; and performing localization by combining SFM, ground plane and object bounding boxes to produce a 3D object localization.

Abstract: Systems and methods are disclosed for autonomous driving with only a single camera by moving object localization in 3D with a real-time framework that harnesses object detection and monocular structure from motion (SFM) through the ground plane estimation; tracking feature points on moving cars a real-time framework to and use the feature points for 3D orientation estimation; and correcting scale drift with ground plane estimation that combines cues from sparse features and dense stereo visual data.

Abstract: Systems and methods are disclosed for multithreaded visual odometry by acquired with a single camera on-board a vehicle; using 2D-3D correspondences for continuous pose estimation; and combining the pose estimation with 2D-2D epipolar search to replenish 3D points.