Sparse 1D Range Data Obstacle Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face computational inefficiencies when navigating by processing dense three-dimensional point clouds to distinguish between obstacles and ground surfaces, especially at high speeds or with low-resolution sensors, which can increase data acquisition and processing times.
Innovation Solution
The method involves using sparse one-dimensional range scan data to determine if points in the scan data are likely to be obstacles or flat ground by comparing them to other points and applying specific geometric constraints, thereby reducing the need for high-resolution data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dense three-dimensional point clouds are processed to distinguish obstacles from ground surfaces, then segmentation accuracy is improved, but computational efficiency deteriorates
Solution Approach 1:
The patent extracts only the necessary ground points from the point cloud data by applying geometric constraints and slope comparisons, rather than processing the entire dense point cloud. This selective extraction approach maintains segmentation accuracy while significantly reducing computational load.
Solution Approach 2:
The patent segments the point cloud data into ground points and non-ground points using a systematic approach based on geometric constraints. By dividing the processing task into discrete comparison steps (slope calculation, threshold comparison), the system achieves accurate segmentation with improved computational efficiency.
2Measurement precision
If high-resolution sensor data is used for environment perception, then detection precision is improved, but data acquisition time increases
Solution Approach 1:
The patent applies partial action by using only the necessary geometric constraints and comparisons needed for ground segmentation, rather than processing all available high-resolution data. This approach maintains sufficient detection precision while reducing data acquisition and processing time.
3Measurement precision
If dense point cloud processing is performed at high vehicle speeds, then segmentation accuracy is maintained, but processing time increases
Solution Approach 1:
The patent extracts ground points using efficient geometric constraints that can be computed rapidly even at high vehicle speeds. By focusing only on the essential comparisons (slope calculations and threshold checks), the system maintains segmentation accuracy while achieving real-time processing performance.
Data Source
AI summary
Methods and systems for object and ground segmentation from a sparse one-dimensional range data are described. A computing device may be configured to receive scan data representing points in an environment of a vehicle. The computing device may be configured to determine if a test point in the scan data is likely to be an obstacle or ground by comparing the point to other points in the scan data to determine if specific constraints are violated. Points that do not pass these tests are likely to be above the ground, and therefore likely belong to obstacles.