Autonomous Driving Map Mesh for Lane-Aware Rendering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems face inefficiencies and redundant memory usage due to the need to separately render drivable areas and lane segments, which leads to computational inefficiencies and map visualization issues.
Innovation Solution
The system segments a drivable area into a mesh of polygons using lane segment boundaries, allowing each lane segment to be represented as a union of a unique subset of sub-regions, thereby combining data into a single geographical layer for efficient rendering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If drivable area and lane segments are rendered separately, then detailed map information can be provided, but computational efficiency decreases and memory usage increases
Solution Approach 1:
The drivable area is segmented into multiple sub-regions based on lane segment boundaries. Each lane segment is represented as a union of specific sub-regions, creating a hierarchical structure that enables efficient rendering by allowing the system to process and prioritize individual lane segments within the larger drivable area context
Solution Approach 2:
The patent combines the drivable area representation with lane segment information by representing lane segments as unions of sub-regions. This merging eliminates redundant data storage while preserving both the overall drivable area context and specific lane segment details needed for accurate mapping and efficient rendering
2Device complexity
If drivable area is represented as large segments, then simplified driving environment representation is achieved, but ability to infer information about discrete sections is lost
Solution Approach 1:
The drivable area is divided into multiple sub-regions that correspond to discrete lane segments. This segmentation maintains the simplified representation benefit while enabling access to detailed lane-specific information through the sub-region structure, where each lane segment can be reconstructed as a union of its constituent sub-regions
Solution Approach 2:
Different sub-regions are assigned different properties based on their association with specific lane segments. This allows the system to maintain uniform data structures while providing localized detailed information about discrete sections when needed, as each sub-region can be selectively processed based on its lane segment associations
3Loss of information
If separate rendering of drivable area and lane segments is performed, then complete map information is provided, but memory usage becomes redundant
Solution Approach 1:
The patent merges the drivable area data structure with lane segment information by representing lane segments as unions of sub-regions. This eliminates redundant storage of lane segment data while maintaining complete map information, as the same sub-region data serves both the overall drivable area representation and specific lane segment definitions
Data Source
AI summary
Methods and systems for controlling navigation of an autonomous vehicle for traversing a geographical area are disclosed. The methods include receiving information relating to a drivable area in the geographical area and identifying a plurality of lane segments that intersect with the drivable area. The plurality of lane segments can be used to segment the drivable area into a plurality of sub-regions such that each of the plurality of lane segments can be represented as a union of a unique subset of the plurality of sub-regions. A data representation of the drivable area may be created to include the plurality of sub-regions, and used to render the map of the geographical area. The map includes the drivable area and/or one or more of the plurality of lane segments.