Road Network Mapping with Lattice Graphs for Autonomous Driving Plans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems face challenges in generating accurate and comprehensive road network data necessary for designing both short-term and long-term driving plans.
Innovation Solution
A method involving a computing device that generates road network data, converts it into lattice road network data for short-term plans, and further processes it to create a crossable dipole graph for long-term plans, utilizing graph generation models and lattice road network structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If road network data is generated for autonomous driving, then navigation capability is improved, but data complexity and processing difficulty increase
Solution Approach 1:
The road network data is segmented into multiple hierarchical levels: road network data (basic connectivity), lattice road network data (added spatial lattice structures), and crossable dipole graphs (added lane-level crossing information). This segmentation allows the system to process and utilize road information in manageable stages, improving navigation capability while controlling data complexity through progressive refinement.
Solution Approach 2:
The patent transforms road network data into lattice road network data by adding a lattice dimension, and further transforms it into crossable dipole graphs by adding lane-level dimensional information. This dimensional transformation enables the system to handle complex navigation scenarios by enriching the data structure without overwhelming processing requirements.
2Measurement precision
If comprehensive road network data is generated, then driving plan accuracy is improved, but processing time increases
Solution Approach 1:
The system performs preliminary processing by generating road network data first, then progressively creates lattice road network data and crossable dipole graphs in advance. This preliminary action ensures that comprehensive road network information is prepared beforehand, improving driving plan accuracy while reducing real-time processing time during actual autonomous driving operations.
Solution Approach 2:
The data processing is divided into distinct segments: road network data generation, lattice road network data generation, and crossable dipole graph generation. Each segment can be processed independently and cached, allowing the system to achieve high driving plan accuracy through comprehensive data while minimizing real-time processing time by reusing pre-generated data structures.
3Manufacturing precision
If detailed lane information is included in road network data, then route planning precision is improved, but data structure complexity increases
Solution Approach 1:
The patent adds lane-level information by transforming road network data into crossable dipole graphs, which incorporate detailed lane connectivity and crossing information. This dimensional enhancement provides high route planning precision by capturing fine-grained lane-level details without requiring a complete redesign of the underlying data structure, as it builds upon the existing road network and lattice road network frameworks.
Solution Approach 2:
The data structure follows a nested hierarchy where crossable dipole graphs (containing lane-level details) are nested within lattice road network data (containing spatial lattice structures), which in turn are nested within road network data (containing basic connectivity). This nested structure allows detailed lane information to be incorporated for high route planning precision while managing data structure complexity through hierarchical organization, enabling efficient access to different levels of detail as needed.
Data Source
AI summary
Provided are a method, a server, and a computer program for creating a road network map to design a driving plan for an autonomous driving vehicle. A method of creating a road network map to design a driving plan for an autonomous driving vehicle is performed by a computing device and includes: generating road network data for an area; generating lattice road network data for a short-term driving plan for an autonomous driving vehicle using the generated road network data; and generating a crossable dipole graph for a long-term driving plan for the autonomous driving vehicle using the generated lattice road network data.