Lane Marking Recognition Reliability via Multi-Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing travel lane marking recognition systems face challenges in accurately determining the reliability of lane markings, leading to excessive restriction or erroneous driving assistance when imaging conditions are poor, such as in backlighting, rain, or fog.
Innovation Solution
A travel lane marking recognition system that incorporates an on-board camera and additional state acquisition apparatuses like stereo cameras, millimeter wave radar, and ultrasonic sensors to capture vehicle and environmental data, allowing for the setting of appropriate reliability levels for lane markings, enabling accurate driving control and deviation warnings based on these levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If driving assistance is performed based on travel lane marking recognition using only on-board camera image data, then the system can provide driving assistance, but the accuracy of reliability calculation decreases when imaging conditions are poor
Solution Approach 1:
The patent combines multiple state acquisition apparatuses (camera, radar, LIDAR, ultrasonic sensor) to acquire different types of data about the travel lane and vehicle state. By merging these diverse data sources, the system can calculate reliability more accurately even when imaging conditions are poor, as the additional sensors provide complementary information that compensates for camera limitations.
Solution Approach 2:
The patent introduces a reliability level setting unit that acts as an intermediary between the raw sensor data and the driving assistance control. This unit processes and evaluates data from multiple sources to determine a reliability level, which then guides whether driving assistance should be activated. This intermediary layer ensures that driving assistance is only provided when sufficient reliable data is available.
2Reliability
If the reliability threshold for driving assistance is set high, then erroneous driving assistance is reduced, but driving assistance is excessively restricted
Solution Approach 1:
The patent implements dynamic threshold adjustment where the reliability threshold for driving assistance is not fixed but adapts based on the accumulated reliability levels from multiple state acquisition apparatuses. When multiple sensors confirm the same travel lane marking, the system dynamically lowers the threshold to allow driving assistance, whereas with single-sensor data, the threshold remains high. This dynamic approach balances reliability and availability.
Solution Approach 2:
The system changes the reliability parameter based on the number and type of state acquisition apparatuses that confirm the travel lane marking. When multiple apparatuses provide consistent data, the system adjusts the reliability parameter upward, enabling driving assistance even in challenging conditions. This parameter change allows the system to overcome the trade-off between reliability and availability.
3Reliability
If multiple state acquisition apparatuses are used to improve reliability assessment, then the system can appropriately set reliability levels, but the device complexity increases
Solution Approach 1:
The patent designs the state acquisition apparatuses to serve multiple functions: they not only detect travel lane markings but also provide information about vehicle state, surrounding environment, and imaging conditions. This multi-functionality allows the system to use the same hardware for multiple purposes, reducing the need for additional dedicated sensors and thereby limiting the increase in device complexity.
Solution Approach 2:
The system implements a modular approach where the full multi-sensor suite is not always active. Instead, the system activates additional state acquisition apparatuses only when needed based on imaging conditions and reliability requirements. This partial action approach allows the system to have high reliability capability available when needed while maintaining lower operational complexity during normal conditions.
Data Source
AI summary
An on-board camera captures an image of a travel lane ahead of an own vehicle. At least one state acquisition apparatus acquires an acquired value indicating a state of the own vehicle. A travel lane marking recognition system recognizes a travel lane marking that demarcates the travel lane from an image captured by the on-board camera. The travel lane marking recognition system sets a reliability level of the recognized travel lane marking based on the acquired value of the at least one state acquisition apparatus. When the reliability level is higher than a first predetermined threshold, the travel lane marking recognition system performs driving control of the own vehicle based on the recognized travel lane marking. When the reliability level is lower than the first predetermined threshold, the travel lane marking recognition system performs a deviation warning of deviation from the travel lane based on the recognized travel lane marking.