Autonomous Vehicle Point Cloud Filtering for 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 hazards.
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 traverse through particulate matter by discarding irrelevant data from the planning process.
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 segments LIDAR point cloud data into multiple blobs or clusters, where each blob represents a potential object or region. By processing and classifying each blob separately, the system can distinguish between actual obstacles and particulate matter more effectively, resolving the contradiction between comprehensive detection and accurate identification.
Solution Approach 2:
The patent introduces an intermediary classification process that acts as a mediator between raw LIDAR detection and obstacle identification. This intermediary layer analyzes blob characteristics (such as reflectivity, size, shape, and spatial distribution) to determine whether detected objects are true obstacles or particulate matter, thereby improving reliability without sacrificing detection coverage.
2Reliability
If all LIDAR data is processed for route planning, then navigation safety is improved, but processing time and computational load increase
Solution Approach 1:
The patent extracts and removes irrelevant data (particulate matter blobs) from the LIDAR point cloud before proceeding with route planning. By taking out only the necessary obstacle information and discarding false detections, the system reduces computational load and processing time while maintaining navigation safety.
Solution Approach 2:
The patent applies partial processing by focusing computational resources only on classified obstacle blobs rather than processing all LIDAR data uniformly. This selective approach processes only the critical subset of data needed for safe navigation, significantly reducing processing time while maintaining safety through targeted analysis of potential obstacles.
3Measurement precision
If LIDAR data includes particulate matter, then detection completeness is improved, but route planning efficiency decreases
Solution Approach 1:
The patent performs preliminary classification of LIDAR blobs into obstacle and non-obstacle categories before the route planning process begins. This preliminary action filters out particulate matter and other non-obstacle elements in advance, so that the subsequent route planning algorithm only needs to consider true obstacles, thereby maintaining detection completeness while improving planning efficiency.
Solution Approach 2:
The patent applies different processing qualities to different regions of the LIDAR data based on their classification. Obstacle blobs receive detailed processing suitable for safe navigation, while non-obstacle blobs (particulate matter) receive minimal or no processing. This local differentiation maintains detection completeness for important objects while improving overall efficiency by reducing unnecessary processing.
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.