Point Cloud Filtering Using Semantic Segmentation in Particulate Matter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating through environments with particulate matter like fog, steam, or dust, as current sensor data often incorrectly identifies these as obstacles, leading to inefficient route planning and potential safety issues.
Innovation Solution
The implementation of a filtering system that uses LIDAR data and image data through semantic segmentation to determine whether a LIDAR blob represents a true obstacle or a navigable space, allowing the vehicle to safely pass through particulate matter by filtering out irrelevant data from trajectory planning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LIDAR data is used to detect all objects in the environment, then detection coverage is improved, but false identification of particulate matter as obstacles increases
Solution Approach 1:
The patent introduces image data from a camera as an intermediary to verify LIDAR detections. The camera captures visual information about particulate matter, and through semantic segmentation, determines whether detected LIDAR blobs correspond to actual obstacles or navigable spaces. This intermediary verification mechanism resolves the contradiction by providing additional context to distinguish true obstacles from false positives like fog or dust clouds.
2Reliability
If all LIDAR data is processed for trajectory planning, then safety is improved, but processing time and computational requirements increase
Solution Approach 1:
The patent extracts and removes irrelevant LIDAR data points that correspond to navigable spaces (particulate matter) after verification by the image processing system. By taking out only the necessary obstacle data for trajectory planning and discarding false positive LIDAR returns, the system maintains safety while significantly reducing processing time and computational load on the autonomous vehicle's planning algorithms.
3Measurement precision
If semantic segmentation is applied to verify LIDAR detections, then obstacle identification accuracy is improved, but data processing requirements increase
Solution Approach 1:
The patent applies semantic segmentation specifically to regions of interest identified by LIDAR detections rather than processing entire images. This targeted approach segments only the relevant portions of image data corresponding to LIDAR-detected blobs, reducing overall computational requirements while maintaining high obstacle classification accuracy. The segmentation is performed selectively on verified detection regions rather than globally.
Data Source
AI summary
Techniques are disclosed for filtering point cloud data associated with particulate matter (e.g., gas, exhaust, fog, etc.) which do not impact driving from data used to plan a trajectory and/or a route of a robotic platform. The filtering may be based on determining that a set of points associated with a point cloud represents a navigable space for the robotic platform. The point cloud data may be filtered based on a determination that one or more safety conditions are satisfied. Semantic segmentation may be performed on an image to determine pixel classification probability distributions associated with pixels of the image. Data associated with the set of points may be projected onto the image to identify corresponding pixels. Confidence scores associated with pixel classification probability distributions for the identified pixels may be queried. A classification probability distribution for the set of points may be determined based at least in part on the queried confidence scores.