Dynamic Map Augmentation for Autonomous Vehicle Lane Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately detecting and managing various types of boundaries, especially in construction areas, where physical barriers, temporary lane markers, and spaced construction elements can diverge from mapped data, leading to difficulties in determining the correct lane and trajectory.
Innovation Solution
An autonomous vehicle control system that uses a unified boundary machine learning model to detect multiple types of boundaries from perception data, integrating detection of perceived boundaries with map data to generate augmented map data, allowing the vehicle to operate in perceived lanes and adapt to discrepancies between mapped and perceived lanes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If high quality digital maps are used to represent static objects and logical elements, then trajectory generation is improved, but map data updates require substantial overhead and may not reflect sudden environmental changes
Solution Approach 1:
The system dynamically switches between relying on pre-stored map data and relying on real-time perception data based on environmental conditions. When construction zones or temporary boundaries are detected, the system transitions from static map-based navigation to dynamic perception-based navigation, allowing adaptive response to changing environments without requiring continuous map updates
Solution Approach 2:
The autonomous vehicle performs self-updating of its environmental model by using its own perception system to detect and interpret current lane boundaries and construction elements. The vehicle independently identifies discrepancies between map data and actual environment, and adjusts its navigation accordingly without external intervention or map distribution overhead
2Reliability
If perception systems are used to detect changed circumstances, then real-time environmental awareness is improved, but detection range is limited and can be occluded by other vehicles
Solution Approach 1:
The system performs preliminary interpretation of perception data by identifying and classifying construction elements and temporary boundaries before they become critical obstacles. By pre-processing and pre-interpreting sensor data to detect potential lane boundary changes, the system prepares navigation adjustments in advance, allowing earlier reaction to environmental changes despite limited detection range
3Measurement precision
If multiple types of boundaries are detected in construction areas, then lane identification accuracy is improved, but system complexity increases
Solution Approach 1:
The system segments boundary detection into distinct functional modules: detecting physical barriers, detecting temporary lane markers, detecting spaced construction elements, and interpreting their combined meaning. Each module handles a specific type of boundary detection independently, and the results are integrated to form a complete lane identification, reducing overall system complexity through functional decomposition
Solution Approach 2:
The system introduces an intermediary interpretation layer that reconciles discrepancies between map data and perceived lane elements. This intermediary component processes detection results from multiple sources, resolves conflicts between different boundary types, and generates unified navigation instructions, simplifying the integration of multiple detection functions
Data Source
AI summary
Discrepancies between mapped lanes on a roadway and perceived lanes perceived by the perception system of an autonomous vehicle may be identified and used to augment map data used by the autonomous vehicle, thereby enabling the autonomous vehicle to operate even in the event of the identified discrepancy.