Object Orientation Determination for Sparse LiDAR Vehicle Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately determining the position and orientation of other vehicles, especially when receiving sparse LiDAR point clouds due to distant objects or partial occlusion, making it difficult to identify the front or back of vehicles, particularly for static vehicles where object motion cannot be used.
Innovation Solution
The method involves obtaining map and group parameters using a processor to determine object orientation, incorporating these parameters into object detection data to improve object detection and tracking, and enhancing neural network models for estimating object orientations, dimensions, and locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR is used to detect objects in the environment, then object detection capability is provided, but detection accuracy deteriorates when point clouds are sparse due to distant objects or partial occlusion
Solution Approach 1:
The patent introduces map parameters and group parameters as intermediary information to bridge the gap between sparse LiDAR data and accurate object orientation determination. These parameters act as mediators that provide additional contextual constraints to resolve the information loss from sparse point clouds
Solution Approach 2:
The system performs preliminary actions by pre-obtaining map parameters (indicative of predetermined object positions) and group parameters (indicative of predetermined relations between objects) before object detection. These pre-acquired parameters are then integrated with LiDAR data to improve detection accuracy when point clouds are sparse
2Measurement precision
If traditional object detection methods are used, then detection process is simple, but orientation determination accuracy deteriorates for static vehicles where motion cannot be utilized
Solution Approach 1:
The patent merges multiple information sources including LiDAR point cloud data, map parameters, and group parameters into a unified object detection framework. This combination allows the system to leverage both motion cues from dynamic objects and contextual constraints from static map and group information to determine orientations accurately
Solution Approach 2:
The system changes the parameter set used for object detection by incorporating map parameters (predetermined positions) and group parameters (predetermined relations) alongside traditional LiDAR features. This parameter expansion enables accurate orientation determination for static vehicles that cannot provide motion-based cues
Data Source
AI summary
Provided are methods for object orientation determination, which can include obtaining map parameters and group parameters and determining orientation data using said map and group parameters. Some methods described also include obtaining sensor data and using the sensor data for the determination of orientation data. Systems and computer program products are also provided.