Local Window 2D Occupancy Grids for Autonomous Vehicle Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 2D occupancy grids for roadways inaccurately represent structures above the road surface, such as bridges and overpasses, leading to false identification of grid cells as occupied, which can hinder accurate localization and mapping applications in autonomous vehicles.
Innovation Solution
Generating 2D occupancy grids using overlapping or stacked windows with a predetermined altitude range for point cloud data, limiting data to only what is within a specific height above the road surface, and determining grid cell occupancy based on a threshold of point cloud data points, thereby excluding data from structures above the road.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all point cloud data is used to generate 2D occupancy grids, then complete environmental representation is achieved, but false positives from overhead structures (bridges, overpasses) contaminate the localization data
Solution Approach 1:
The patent applies local quality by creating multiple local windows with different altitude ranges tailored to specific geographic locations. Each window is configured with altitude thresholds appropriate for its local environment (e.g., excluding overhead structures in urban areas while capturing all relevant data in open areas), allowing the system to maintain high localization accuracy without false positives while preserving necessary environmental information for each specific context
Solution Approach 2:
The patent segments the environmental data processing into multiple local windows, each handling a specific geographic region with its own altitude filtering parameters. This segmentation allows different parts of the environment to be processed with locally-optimized settings, resolving the contradiction between maintaining complete environmental representation and excluding contaminating overhead structures
2Measurement precision
If a large altitude range is used for point cloud data, then complete roadside object detection is achieved, but computational and networking resources are excessively consumed
Solution Approach 1:
The patent dynamically adjusts the altitude range parameter based on the specific local window and geographic context. By changing this parameter locally rather than using a fixed large range everywhere, the system maintains precise roadside object detection where needed while reducing computational and networking resource consumption in areas where full altitude coverage is unnecessary
Solution Approach 2:
The patent applies partial action by using only the necessary portion of the altitude range for each local window. Instead of processing all possible altitude data universally, the system processes only the partial range required for each specific location, achieving sufficient detection accuracy while minimizing resource consumption
3Measurement precision
If high resolution 2D occupancy grids are generated for entire roadway regions, then precise localization is achieved, but data processing and storage requirements become unmanageable
Solution Approach 1:
The patent divides the large roadway region into multiple smaller local windows, each generating its own high-resolution 2D occupancy grid. This segmentation allows the system to maintain precise localization within each window while managing data quantities at a scale that is computationally feasible and storage-efficient
Solution Approach 2:
The patent transitions from generating a single large 2D occupancy grid for the entire roadway region to generating multiple smaller local windows with their own 2D grids. This dimensional reorganization (from one large grid to many small grids) maintains localization precision within each window while dramatically reducing the total data burden through manageable partitioning
Data Source
AI summary
Embodiments include apparatus and methods for automatic generation of local window-based 2D occupancy grids that represent roadside objects at a region of a roadway and automatic localization based on the 2D occupancy grids. 2D occupancy grids are generated based on an altitude threshold for point cloud data and grid cell occupancy for grid cells of local windows associated with the region of the roadway. The 2D occupancy grids are stored in a database and associated with the region of the roadway. Sensor data from a user located at the region of the roadway is received. The accessed 2D occupancy grids and the received sensor data are compared. Based on the comparison, localization of the user located at the region of the roadway is performed.