Stereo Ground Surface Estimation for Sparse LiDAR Road Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques for estimating road surfaces in autonomous vehicles suffer from limited accuracy, particularly due to the sparsity of LiDAR data at greater distances and the inability of RADAR and camera-only solutions to provide precise distance measurements, leading to poor handling on uneven surfaces and potential vehicle instability.
Innovation Solution
The method involves using nonlinear optimization to fit height values to bias-corrected LiDAR detections, combined with stereo imaging to generate a disparity field, which enhances the accuracy of road surface estimation by compensating for measurement biases and refining disparity values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR is used for road surface estimation, then measurement precision is improved at close range, but data sparsity occurs at greater distances
Solution Approach 1:
The patent combines LiDAR data with stereo camera imagery to create a fused representation of the road surface. The LiDAR provides precise depth measurements while the stereo camera provides dense visual coverage, and their integration compensates for the sparsity of LiDAR data at distance through data fusion techniques.
Solution Approach 2:
The patent introduces an intermediate processing stage that uses stereo disparity fields as a mediator between the sparse LiDAR measurements and the final road surface model. The disparity field provides additional geometric constraints and visual context that fill in the gaps where LiDAR data is sparse.
2Length of stationary object
If RADAR is used for long range detection, then detection range is improved, but measurement precision deteriorates
Solution Approach 1:
The patent merges RADAR's long-range detection capability with the precision of LiDAR and stereo vision. The system uses RADAR for initial long-range object detection and tracking, then integrates this with LiDAR and camera data for precise road surface modeling, allowing the system to leverage the strengths of each sensor modality.
3Illumination intensity
If camera-only solutions are used, then image resolution is improved, but distance estimation precision deteriorates
Solution Approach 1:
The patent replaces pure visual (camera-based) depth estimation with a hybrid approach that incorporates active sensing from LiDAR. Instead of relying solely on passive camera imaging and monocular depth cues, the system uses LiDAR's time-of-flight or phase-shift measurements to obtain direct, precise depth information that complements the high-resolution camera imagery.
4Length of stationary object
If LiDAR data is used for far distance measurement, then detection capability is improved, but measurement accuracy deteriorates due to weather conditions
Solution Approach 1:
The patent creates a composite sensing system that combines multiple sensor types (LiDAR, stereo cameras, and potentially RADAR) to achieve robust far-distance measurement. Each sensor modality has different weather dependencies, and their combination creates a more reliable system where one sensor can compensate for another's weaknesses in adverse weather conditions.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves the precision of road surface estimation, enabling safer and more comfortable autonomous vehicle operation by accurately detecting obstacles and navigating uneven terrain.
Implementation Method 1
LiDAR, which uses laser pulses to create detailed 3D representations of the environment
Implementation Method 2
RADAR, which uses radio waves to detect objects and measure distances
Implementation Method 3
stereo imaging to generate a disparity field
Data Source
AI summary
Embodiments of the present disclosure relate to surface estimation using stereo imaging and surface disparities. For example, a three-dimensional (3D) surface structure may be modeled as a disparity field, and a surface disparity field representing a surface in the environment (e.g., the ground) may be generated using a constrained nonlinear hierarchical optimization to process stereo image data and iteratively refine estimated surface disparity values based on weights that guide the optimization to expected surface values (e.g., ground, road). The resulting surface (e.g., ground) disparity field may be used for a variety of downstream tasks, such as obstacle detection, segmentation of a navigable space, ego-motion refinement, and/or generation of an estimated surface profile.