Roundabout Path Recognition Using Vehicle Trajectory Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face challenges in detecting road pathways in roundabout traffic circles due to high curvature lanes, blocked lane marks by nearby vehicles, and central islands, which limit the effectiveness of optical sensors like cameras and lidars.
Innovation Solution
The implementation of a roundabout characteristics recognition system utilizing Random Finite Set (RFS) modules, which process raw data from sensors to extract feature characteristics, determine the reliability of detected trajectories, and generate accurate topological and geometric data for path planning and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical sensors (camera or lidar) are used to detect lane marks, then high-accuracy guidance is achieved under moderate traffic density, but detection reliability deteriorates in roundabout environments with high curvature lanes and blocked lane marks
Solution Approach 1:
The patent introduces trajectory information from remote vehicles as an intermediary data source to supplement direct optical sensor detection. By using the moving trajectories of other vehicles as mediators to infer lane information, the system overcomes the limitation of blocked lane marks in roundabouts, maintaining detection reliability when direct visual detection fails
Solution Approach 2:
The system makes the detection system universal by enabling it to function effectively in both moderate traffic conditions (using direct optical detection) and roundabout environments (using trajectory-based inference). The multi-functional detection approach allows the same system to adapt to different environmental conditions and traffic patterns, achieving reliable lane detection across diverse scenarios
2Adaptability or versatility
If trajectory information from remote vehicles is used to detect roundabout pathways, then detection capability is enhanced beyond pavement detection, but system complexity increases due to multiple RFS modules and statistical processing
Solution Approach 1:
The patent segments the complex detection task into multiple specialized Random Finite Set (RFS) modules, each handling specific aspects of trajectory analysis. By dividing the overall system into modular components that process different types of data and perform specific functions, the system manages complexity through structured organization while maintaining enhanced detection capabilities
Solution Approach 2:
The patent replaces traditional mechanical/optical detection methods with statistical processing and probability density function analysis. Instead of relying solely on physical sensor detection, the system uses mathematical models and statistical inference to extract lane information from trajectory data, substituting computational methods for direct physical measurement
Data Source
AI summary
A roundabout characteristics recognition system for an autonomous driving vehicle. The roundabout characteristics recognition system detects the various remote object trajectories around the roundabout traffic circle environment and analyzes the trajectories to extract feature data relevant to the roundabout topology and geometry. The roundabout characteristics recognition system processes the remote object trajectories data and determines the statistical characteristics of the various topological characteristics and geometric characteristics for path planning and control to make a more precise decision for operation of the autonomous driving vehicle.