Autonomous Vehicle Trajectory Inference for Unreliable Lane Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating when lane information becomes unavailable or unreliable, such as due to errors in maps or missing lane markings, which can lead to collisions with other vehicles or obstacles.
Innovation Solution
The vehicle's computer system identifies trajectories of other vehicles to locate potential merge points, determines the lane location at these points, creates a new trajectory, and controls the vehicle to travel along this new path, ensuring safe navigation even without reliable lane information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If autonomous vehicles rely on traditional lane information (maps and lane markings) for navigation, then navigation accuracy is improved, but the system becomes vulnerable to collisions when lane information is unavailable or unreliable
Solution Approach 1:
The patent introduces other vehicles as intermediary objects to infer lane information. When traditional lane markings or map data become unavailable, the system uses trajectories and positions of surrounding vehicles as proxy indicators to determine the current lane location, thereby maintaining navigation reliability without sacrificing accuracy
Solution Approach 2:
The system creates a virtual copy of lane information by inferring lane boundaries from the collective trajectories of multiple other vehicles. This copied lane data allows the autonomous vehicle to navigate as if traditional lane markings were present, resolving the contradiction between relying on accurate lane information and maintaining reliability when such information is absent
2Measurement precision
If the vehicle uses sensors to detect lane markings and map data to determine lane location, then lane tracking precision is improved, but the system fails when lane markings are missing or map data is erroneous
Solution Approach 1:
The system makes the lane detection mechanism universal by enabling it to function through multiple data sources: traditional lane markings, map data, and inferred trajectories from other vehicles. This multi-functionality allows the system to adapt to various road conditions including missing lane markings or erroneous map data while maintaining detection precision
Solution Approach 2:
The system performs preliminary analysis of other vehicle trajectories to pre-establish alternative lane location references before traditional lane detection fails. By continuously monitoring and storing trajectory data, the system prepares backup lane information that can be immediately activated when lane markings disappear or map data proves incorrect
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Methods and devices for controlling a vehicle in an autonomous mode are disclosed. In one aspect, an example method is disclosed that includes obtaining, by a computer system, lane information that provides an estimated location of a lane of a road on which a vehicle is travelling, where the computer system is configured to control the vehicle in an autonomous mode. The example method further includes determining, by the computer system, that the lane information has become unavailable or unreliable and, in response to determining that the lane information has become unavailable or unreliable, the computer system analyzing trajectories of other vehicles to locate a potential merge point on the road and creating a new trajectory that follows the lane at the potential merge point.