Lane-Merge Object Recognition for Autonomous Driving Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle object recognition systems struggle to accurately identify the merging of lanes and classify objects in these areas, leading to potential misidentification of moving and stationary objects, which can cause sudden braking issues and increase the risk of accidents.
Innovation Solution
An object recognition apparatus and method that utilizes a sensor, such as LIDAR, to determine the lateral and longitudinal positions of road boundaries and structures, assign reliability values based on object size and shape, and generate control signals for autonomous driving to manage merging sections, thereby improving object classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If the vehicle uses sensor data to detect objects in merging lane sections, then the ability to identify objects is improved, but the accuracy of distinguishing moving objects from stationary objects deteriorates due to the complex merging scenario
Solution Approach 1:
The system segments the road into different sections based on lane merging detection. When a merging section is detected through road boundary analysis, the system applies different object classification rules specifically for that segment. This allows the system to handle the complex merging scenario with specialized logic while maintaining standard detection elsewhere.
Solution Approach 2:
The system performs preliminary detection of road boundaries and lane merging sections before object classification. By first identifying the merging section through road structure analysis, the system prepares the appropriate classification context in advance, improving the accuracy of distinguishing moving versus stationary objects in the merging area.
2Object-affected harmful factors
If the system classifies objects as stationary in merging sections, then false positive braking is reduced, but the risk of missing actual moving objects increases
Solution Approach 1:
The system dynamically adjusts object classification based on the detected road section. In merging sections, the system uses a two-stage approach: first assuming stationary objects to reduce false positives, then verifying through reliability assessment whether the object is truly stationary or a moving object that should trigger braking.
Solution Approach 2:
The system implements feedback through reliability value assessment. After initially classifying objects in merging sections, the system evaluates the reliability of this classification by analyzing object characteristics, sensor data consistency, and contextual information. This feedback loop ensures that potentially dangerous moving objects are not mistakenly classified as stationary.
3Reliability
If the vehicle increases caution in merging sections by treating all objects as potential hazards, then safety is improved, but sudden braking occurs more frequently affecting driving smoothness
Solution Approach 1:
The system applies different safety levels to different spatial locations. In merging sections, the system enhances safety through careful object classification and reliability assessment, while in non-merging sections, normal driving operations continue without excessive caution. This localized approach ensures safety where needed without disrupting overall driving smoothness.
4Measurement precision
If the system uses multiple sensor inputs and analysis methods to improve object classification, then classification accuracy is improved, but the system complexity increases
Solution Approach 1:
The system segments the processing logic into distinct modules: road boundary detection, merging section identification, object detection, and classification. Each module handles a specific aspect of the problem independently. This modular segmentation improves classification accuracy through comprehensive analysis while managing system complexity through organized, separable processing stages.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances the accuracy of object classification in merging lane scenarios, reducing the risk of misidentification and potential accidents by providing precise control signals for autonomous vehicles.
Implementation Method 1
A distance from a LIDAR to an object may be obtained through an interval between the time when laser is transmitted by the LIDAR and the time when the laser reflected by the object is received
Data Source
AI summary
An apparatus for controlling autonomous driving of a vehicle is introduced. The apparatus may comprise a sensor and a processor configured to determine, based on the sensor sensing at least one of a lateral position of a point on a road boundary or a lateral position of at least one road structure, a merging of at least two different lanes, determine, based on at least one of a time period associated with the merging or a distance associated with the merging, a merging section, determine a type of an object located in the merging section, wherein the type is one of a moving object, a first stationary object that is able move, or a second stationary object that is unable move, and generate, based on the type, a control signal for controlling the autonomous driving of the vehicle in the merging section.