3D LiDAR Intensity Descriptors for Fast Global Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for global localization in 3D environments, particularly in mobile robotics and autonomous vehicle navigation, are inefficient and unsuitable for real-time applications, especially in unstructured environments without GPS infrastructure and under varying lighting conditions, as they rely on computationally complex geometrical recognition or limited global descriptors.
Innovation Solution
A method using intensity descriptors calculated from laser sensor data, dividing the local point cloud into spatially distributed segments, comparing these descriptors with pre-calculated map descriptors to determine location, and refining with geometrical recognition for accurate pose estimation, allowing for efficient localization in unstructured environments and adverse lighting conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If global descriptors are used for place recognition, then dimensionality reduction and correspondence search efficiency are improved, but the ability to determine relative transformation and pose is lost
Solution Approach 1:
The patent segments the point cloud into multiple local regions and computes local descriptors for each region. This segmentation allows the system to maintain global efficiency through histogram-based comparison while preserving local geometric information necessary for accurate pose estimation through subsequent refinement steps.
Solution Approach 2:
The patent transitions from purely global histogram descriptors to a multi-dimensional approach that combines global place recognition with local geometric refinement. This dimensional expansion enables the system to operate efficiently at the global level while achieving precise pose estimation through local geometric analysis.
2Measurement precision
If local geometrical descriptors are used for accurate pose determination, then pose estimation accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent performs preliminary place recognition using computationally efficient global histogram descriptors before engaging in more complex local geometric analysis. This preliminary action filters the search space and identifies candidate locations, allowing subsequent detailed pose estimation to focus only on relevant regions and reducing overall computational complexity.
Solution Approach 2:
The patent applies local geometric descriptor computation selectively rather than uniformly across the entire point cloud. By computing detailed local descriptors only for candidate regions identified through global histogram matching, the system achieves accurate pose estimation where needed while avoiding unnecessary computational overhead in other areas.
3Reliability
If recursive estimation methods like Kalman Filter are used for localisation, then localisation capability is achieved in noisy sensor environments, but the robot must drive around and gather data over time, increasing loss of time
Solution Approach 1:
The patent replaces the mechanical approach of driving around to gather data with a computational approach using intensity-based place recognition. By utilizing the intensity information inherently captured by LiDAR sensors during normal operation, the system achieves reliable localization without requiring additional mechanical movement or time-consuming data gathering maneuvers.
Solution Approach 2:
The patent enables the system to localize itself using intensity information that is already being captured by the LiDAR sensor during normal environmental scanning. This self-service approach eliminates the need for dedicated localization maneuvers or additional sensor infrastructure, as the system utilizes its existing sensing capabilities for dual purposes: environmental mapping and self-localization.
4Loss of information
If cameras are used for place recognition, then visual information can be processed, but performance degrades in unfavourable lighting conditions and darkness
Solution Approach 1:
The patent substitutes optical camera-based recognition with LiDAR-based intensity measurement. This substitution replaces the mechanical/optical system that is sensitive to lighting conditions with a laser ranging system that actively illuminates the environment and measures reflected intensity, providing reliable localization information independent of ambient lighting conditions.
Solution Approach 2:
The patent changes the fundamental parameter used for place recognition from visual intensity (camera) to laser return intensity (LiDAR). This parameter change allows the system to operate in complete darkness because LiDAR carries its own illumination source, making the intensity measurements independent of environmental lighting conditions while providing similar discriminative power for place recognition.
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 enables quick and efficient global localization using a LiDAR sensor alone, reducing computational complexity and achieving accurate pose estimation without external sensors, suitable for indoor and dynamic environments, and can localize a stationary robot without additional data gathering.
Implementation Method 1
determining from a local scan performed by at least one laser sensor, intensity data based at least in part on a power of radiation returned to the at least one laser sensor from points in a local point cloud
Data Source
AI summary
A method for use in performing localisation in a three-dimensional (3D) environment, the method including in one or more electronic processing devices: determining from a local scan performed by at least one laser sensor, intensity data based at least in part on a power of radiation returned to the at least one laser sensor from points in a local point cloud obtained from the local scan; calculating a first intensity descriptor for the local point cloud using the intensity data; retrieving a plurality of previously calculated second intensity descriptors that are each associated with a respective portion of a map of the 3D environment; comparing the first intensity descriptor with at least some of the second intensity descriptors; and, determining a location with respect to the map at least in part in accordance with results of the comparison.