Traffic Flow Detection for ADAS Without Lane Markings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous driving and advanced driver assistance systems (ADAS) rely heavily on lane markings for navigation, which are unreliable in scenarios without markings, poor visibility, or complex road conditions, limiting their ability to accurately determine traffic flow and identify critical road users.
Innovation Solution
A computer-implemented method that detects vehicles in the surrounding environment using sensor data to determine traffic flow characteristics, such as road paths, convoys, and diverging vehicles, without relying on lane markings, by analyzing the pose and motion of detected vehicles to adjust ADAS control functionalities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lane markings are used for navigation and traffic flow determination, then the system can provide a reference for autonomous driving maneuvers, but the system fails in scenarios without lane markings, poor visibility, or complex road conditions
Solution Approach 1:
The patent uses detected vehicles as intermediary objects to infer traffic flow characteristics and road paths. Instead of directly relying on lane markings, the system observes the positions and motions of multiple vehicles to determine convex hulls that represent traffic flow boundaries, enabling navigation in environments where lane markings are unavailable or unreliable
Solution Approach 2:
The patent replaces the visual detection mechanism (camera-based lane marking recognition) with a motion-based inference mechanism. By tracking the positions and velocities of multiple vehicles over time and computing convex hulls of their collective positions, the system derives traffic flow characteristics without requiring visual lane markings
2Measurement precision
If camera-based perception systems are used to detect lane markings, then the system can identify road lanes, but the detection range is limited by sensor resolution and visibility conditions
Solution Approach 1:
The patent combines data from multiple vehicles detected by the sensor system to form a collective representation of traffic flow. By merging the positions of multiple vehicles into a convex hull, the system extends the effective detection range beyond what a single sensor could achieve, as vehicles at different distances contribute to the overall traffic flow characterization
3Difficulty of detecting and measuring
If the system relies on individual vehicle position and motion relative to lane markings, then it can identify critical road users, but it cannot accurately determine traffic flow in the absence of lane markings
Solution Approach 1:
The patent creates a multi-functional approach where the same vehicle detection data used for identifying critical road users is also used to determine overall traffic flow characteristics. The convex hull computation serves multiple purposes: defining traffic flow boundaries, identifying following versus diverging vehicles, and characterizing road paths, thereby preventing loss of traffic flow information
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
A computer-implemented method for determining at least one traffic flow characteristic around a host vehicle, the method comprising: detecting a plurality of vehicles in a surrounding environment of the host vehicle from data acquired from at least one sensor system of the host vehicle and/or at least one sensor system located in the surrounding environment of the host vehicle; determining from the acquired sensor data a set of parameters for each of the detected vehicles comprising a parameter indicative of the pose of the detected vehicle and a parameter indicative of the motion of the detected vehicle; determining the at least one traffic flow characteristic based on the set of parameters obtained for the plurality of vehicles; and determining, based on the determined at least one traffic flow characteristic, an adjustment of at least one control functionality of an advanced driving assistance system.