Autonomous Vehicle Route Sampling Without Precise Map Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving technologies face challenges in generating driving routes when precise map information is unavailable, making it difficult to navigate effectively.
Innovation Solution
A method for generating driving routes using state information of a moving object, including variable and fixed information, to create candidate sample points that satisfy azimuth and separation distance conditions, and selecting optimal sample points based on scoring and free space region conditions to control autonomous driving.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If precise map information is used for route generation, then route accuracy is improved, but system adaptability deteriorates
Solution Approach 1:
The patent introduces an intermediary sampling-based route generation system that mediates between the need for accurate routes and the lack of precise map information. The system uses sampled points and probabilistic road networks as intermediaries to translate incomplete map data into reliable routing decisions, allowing the system to adapt to varying map quality without sacrificing route accuracy
Solution Approach 2:
The system dynamically changes parameters such as sampling density, search radius, and probability thresholds based on the quality and availability of map information. When map information is imprecise or incomplete, the system adjusts these parameters to compensate, maintaining route accuracy while adapting to different environmental conditions
2Measurement precision
If sampling density is increased to improve route precision, then computational complexity increases
Solution Approach 1:
The patent implements dynamic sampling where the sampling density is not fixed but adapts based on the local environment and route requirements. The system increases sampling density in critical areas (such as near obstacles or intersections) while reducing it in open areas, thereby maintaining route precision without uniformly increasing computational complexity across the entire workspace
Solution Approach 2:
The sampling process is segmented into multiple phases or layers, with different sampling densities applied to different regions or stages of route generation. This allows the system to focus computational resources on critical path segments while using coarser sampling for less critical areas, balancing precision and complexity
Data Source
AI summary
A method performed by a vehicle apparatus may comprise generating a plurality of first candidate sample points based on state information of a moving object. These sample points satisfy first azimuth conditions and a separation distance condition, set according to the moving object's state information. The method may further comprise selecting a first optimal sample point from these candidates based on an optimal sample point condition, goal point information, and the candidate points' information. Subsequently, a plurality of second candidate sample points may be generated, satisfying second azimuth conditions and the separation distance condition, using updated state information derived from the first optimal sample point. A second optimal sample point may be then selected from these candidates. Finally, a driving route may be generated based on both the first and second optimal sample points, and the vehicle may be controlled for autonomous driving based on this driving route.