Autonomous Driving Lane Configuration via Trajectory Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles lack sufficient lane configuration information for navigation, as standard navigation maps do not provide detailed lane data, and high-definition maps are expensive to generate and maintain.
Innovation Solution
A system that collects and analyzes perception data from multiple autonomous vehicles to determine lane reference lines and widths, storing this information in a lane configuration database, allowing autonomous vehicles to navigate without relying on high-definition maps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a standard navigation map is used for autonomous driving, then the system complexity is reduced and ease of operation is improved, but the manufacturing precision and measurement precision of lane configuration information deteriorate
Solution Approach 1:
The patent creates a virtual copy of the physical road lane configuration by collecting trajectory data from multiple vehicles and generating virtual lane reference lines that replicate the actual lane structures. This virtual map copy provides sufficient precision for autonomous driving without requiring expensive physical HD map production
Solution Approach 2:
The system enables autonomous vehicles to self-generate lane configuration information by collecting and analyzing their own trajectory data along with data from other vehicles. The vehicles contribute to creating the lane configuration database through their normal operation, eliminating the need for external HD map creation services
2Measurement precision
If a high-definition map is used to provide detailed lane configuration information, then the measurement precision and reliability of lane data are improved, but the cost and device complexity increase significantly
Solution Approach 1:
Instead of using expensive physical HD maps, the system creates a virtual copy of lane configuration data by processing trajectory information from multiple vehicles. This virtual representation provides equivalent measurement precision for autonomous navigation while avoiding the complexity of HD map infrastructure
Solution Approach 2:
The patent introduces trajectory data as an intermediary medium between the physical road environment and the autonomous vehicle's navigation system. By using trajectory data from multiple vehicles as an intermediate representation, the system achieves accurate lane configuration without directly implementing complex HD map technologies
3Reliability
If a high-definition map is generated and maintained, then the reliability of lane configuration information is improved, but the loss of time and resources for map maintenance increase
Solution Approach 1:
The system maintains continuous collection of trajectory data from operating vehicles, which continuously updates and refines the lane configuration database. This continuous data collection process ensures reliability without requiring periodic time-consuming HD map maintenance cycles
Solution Approach 2:
The lane configuration database maintains itself automatically through continuous ingestion of trajectory data from multiple vehicles. The system self-updates lane information based on accumulated operational data without requiring external intervention or dedicated maintenance time
4Ease of manufacture
If standard navigation maps are used without lane configuration details, then the ease of manufacture and device complexity are reduced, but the measurement precision and reliability of navigation information deteriorate
Solution Approach 1:
The system creates a precise virtual copy of lane configurations by processing trajectory data, providing navigation precision equivalent to HD maps while maintaining the ease of using standard navigation map infrastructure
Solution Approach 2:
The patent adds a new dimension of trajectory-based lane configuration information to the traditional two-dimensional navigation map. By incorporating temporal trajectory data from multiple vehicles, the system enriches the map data structure without complicating the underlying standard map infrastructure
Data Source
AI summary
In one embodiment, perception data is received from a number of autonomous driving vehicles (ADVs) over a network. The perception data includes information describing a set of trajectories that a number of vehicles having driven through a road segment of a road and perceived by one or more ADVs using their respective sensors driving on the same road segment. In response to the perception data, an analysis is performed on the perception data, i.e., the trajectories, to determine one or more lanes within the road segment. For each of the lanes, a lane reference line associated with the lane is calculated based on the trajectories within the corresponding lane. The lane metadata describing the lane reference lines for the one or more lanes are stored in a lane configuration data structure such as a database. The lane configuration database can then be utilized for autonomous driving at real-time subsequently without having to use a high definition map.