Intersection Vehicle Detection Using Map Lane Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for object detection and tracking in autonomous vehicles lack the range, processing speed, and high-level accuracy necessary for safe operation at intersections, particularly when combining camera images, lidar, or radar data.
Innovation Solution
A system that combines robust detection algorithms, persistent tracking algorithms, and map data including lane information to accurately detect the position of objects by projecting image data onto map data to determine candidate positions and calculate the actual position of detected objects with minimal error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing object detection methods are used, then the system can detect objects, but the accuracy and processing speed are insufficient for safe autonomous operation at intersections
Solution Approach 1:
The patent segments the object detection process into multiple specialized components: detection algorithms for initial identification, tracking algorithms for persistent monitoring, and map data integration for position verification. This segmentation allows each component to be optimized independently, improving overall accuracy without proportionally increasing processing time.
Solution Approach 2:
The system performs preliminary actions by pre-processing image data to identify candidate objects before full analysis, and by using persistent tracking to maintain object information across frames. This preliminary processing reduces the computational burden on subsequent accuracy-critical steps, enabling both speed and precision.
2Reliability
If existing object detection methods are used, then the system can process data, but the range and accuracy are insufficient for intersection operations
Solution Approach 1:
The patent merges multiple data sources and processing methods into a unified detection system. It combines detection algorithms, tracking algorithms, and map data (including lane information) to cross-validate object positions and improve reliability. This integration increases system complexity but is necessary to achieve the required detection reliability for safe autonomous operation at intersections.
Solution Approach 2:
The system introduces map data as an intermediary layer between raw sensor data and final object position determination. Map data including lane information serves as a reference framework that mediates between camera images/lidar/radar data and the actual road geometry, improving detection reliability by providing contextual constraints without requiring direct modification of the sensor systems.
3Measurement precision
If map data integration is implemented, then the position accuracy improves, but the computational cost increases
Solution Approach 1:
The patent applies local quality by using map data selectively - only integrating map information in regions where objects are detected or where the vehicle is currently located. Rather than processing entire map datasets globally, the system focuses computational resources on local areas of interest, improving position accuracy where needed while reducing overall computational energy consumption.
Data Source
AI summary
A system, device, and methods for image and map-based detection of vehicles at intersections. Once example computer-implemented method for detecting objects includes receiving, from the one or more sensors disposed on a vehicle, image data representative of an image and detecting an object on the image. The method further includes identifying a path extending from the vehicle to the detected object on the image and retrieving map data including lane information. The method further includes comparing the path to a representation of the lane information and determining the position of the detected object based on a comparison of the path, representation of the lane information, and the image.