Real-time Map Generation for Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face difficulties navigating in areas without available high-definition maps, such as rural or new development regions, as they rely on guiding information like lane boundaries and road signs, which are not always present.
Innovation Solution
The generation of real-time maps based on prior driving path data from manned vehicles, using a navigation guideline derived from multiple vehicle trajectories, allowing autonomous vehicles to determine lane boundaries and paths without relying on preconfigured maps, by converting navigation guidelines into body coordinates and dynamically generating maps that can be discarded after use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous vehicles rely on pre-configured high-definition maps for navigation, then navigation accuracy is improved, but the system cannot operate in areas without available maps such as rural or new development regions
Solution Approach 1:
The system pre-processes and stores navigation guidelines and lane boundary data from multiple vehicle trajectories before they are needed. When a vehicle enters an area without HD maps, these pre-computed guidelines are immediately available for navigation, eliminating the need for real-time map generation and enabling operation in previously unmapped regions
Solution Approach 2:
The patent introduces navigation guidelines and lane boundary data as intermediary elements between the vehicle's navigation system and the physical road environment. These intermediaries are derived from collective vehicle trajectory data and serve as a substitute for traditional HD maps, enabling navigation in areas where conventional maps are unavailable
2Adaptability or versatility
If real-time maps are generated for every road segment, then navigation capability in unmapped areas is improved, but storage requirements and processing complexity increase
Solution Approach 1:
The system extracts only the essential navigation elements (navigation guidelines and lane boundary data) from raw vehicle trajectory data, discarding redundant information. This extraction process creates compact, purpose-specific data structures that reduce storage requirements and simplify processing compared to generating complete real-time maps
Solution Approach 2:
Instead of generating complete high-definition maps with all possible road features, the system performs partial action by creating simplified navigation guidelines and lane boundary data containing only the critical information needed for navigation in unmapped areas, thereby reducing computational complexity and storage needs
Data Source
AI summary
In one embodiment, a real time map can be generated by an autonomous driving vehicle (ADV) based on a navigation guideline for a lane and associated lane boundaries for the lane on a particular segment of a road. When travelling in the lane on the road segment, the ADV can use the navigation guideline as a reference line and use the lane boundaries as boundaries. The navigation guideline can be derived from manual driving path data collected by a manned vehicle that has travelled multiple times on the particular segment of the road.