Lane Geometry Estimation via Road Agent Lane Shifts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In vehicular navigation, the absence, outdatedness, or error in map data often leads to difficulties in accurately estimating lane geometry, which is crucial for generating accurate maps based on sensor data.
Innovation Solution
A system comprising sensors and processors that detect road agents and their lane shifts, using this data to estimate boundary lines between lanes by tracking trajectory changes and deviations, employing machine-vision techniques and machine-learning methods like CNNs and LSTM networks to refine lane geometry estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If map data is used for lane geometry estimation, then navigation accuracy is improved, but reliability deteriorates when map data is unavailable, outdated, or erroneous
Solution Approach 1:
The system uses road agents (other vehicles) as intermediary objects to infer lane boundary information. By detecting and tracking the positions and lane-changing behaviors of these intermediary road agents, the system can estimate lane geometry without directly relying on potentially erroneous map data, thus resolving the contradiction between using map data for accuracy and maintaining reliability when map data fails
Solution Approach 2:
The system performs self-service by generating its own lane geometry estimates through sensor-based detection of road agents and their behaviors. Instead of passively relying on external map data that may be outdated or erroneous, the system actively constructs lane geometry information from real-time observations of road agents' positions and lane-changing maneuvers, ensuring reliability independent of map data quality
2Measurement precision
If sensor data from multiple sources is integrated to improve lane geometry estimation, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system segments the lane geometry estimation task into distinct functional modules: road agent detection, lane shift detection, and boundary line estimation. Each module processes specific aspects of the problem using dedicated algorithms (e.g., CNNs for detection, LSTM for trajectory prediction), allowing complex multi-sensor integration to be managed through modular processing stages that improve precision while controlling system complexity
Data Source
AI summary
Systems and methods for estimating lane geometry are disclosed herein. One embodiment receives sensor data from one or more sensors; detects a road agent based on the sensor data; detects, based on the sensor data, that the road agent has performed a lane shift from a first lane of a roadway to a second lane of the roadway; and estimates a boundary line between the first lane of the roadway and the second lane of the roadway based, at least in part, on the detected lane shift.