Vehicle Object Validation Using LiDAR Bounding Box Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems struggle to accurately distinguish between valid and invalid objects, leading to potential collisions and unstable driving conditions, particularly in autonomous driving modes.
Innovation Solution
A vehicle control apparatus using LiDAR sensors and a processor to generate bounding boxes based on contour points and dynamics information, adjusting and aligning these boxes to determine if objects are ghost objects, and employing a neural network model to confirm the classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensors (LiDAR, camera, RADAR) are used to identify external objects, then the detection capability is improved, but the complexity of the system increases
Solution Approach 1:
The patent combines data from multiple sensors (LiDAR, camera, RADAR) into a unified external object identification system. The processor integrates sensor data to generate comprehensive bounding boxes and determine object validity, merging multiple detection sources into a coordinated system that improves accuracy while managing complexity through integrated processing.
2Reliability
If the vehicle braking system responds to all detected external objects, then collision prevention is improved, but driving comfort deteriorates due to sudden braking
Solution Approach 1:
The patent applies different response strategies to different types of detected objects. Valid objects (actual obstacles) trigger collision prevention actions, while invalid objects (ghost objects) are identified and excluded from triggering braking responses. This local differentiation in response quality allows the system to maintain high reliability for real threats while avoiding unnecessary braking that would compromise comfort.
Solution Approach 2:
The patent introduces an intermediary validation process between object detection and braking response. The processor acts as a mediator that analyzes bounding box characteristics, object dynamics, and spatial relationships to determine object validity before triggering braking actions. This intermediary layer filters out false positives while maintaining responsive braking for genuine threats.
3Area of stationary object
If the bounding box is generated based on all contour points, then the object shape coverage is improved, but the accuracy in distinguishing valid and invalid objects deteriorates
Solution Approach 1:
The patent segments the object detection process into multiple stages: generating an initial bounding box covering all contour points, then performing secondary analysis by rotating and regenerating bounding boxes at different angles. This segmentation allows the system to maintain comprehensive coverage while improving discrimination accuracy through multi-angle validation and characteristic analysis of objects within the bounded regions.
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 driving stability and comfort by accurately identifying ghost objects, preventing sudden braking and ensuring safe navigation by distinguishing between valid and invalid objects.
Implementation Method 1
an external object (e.g., debris, vehicles, medians, etc.) may be identified by using various sensors (e.g., light detection and ranging (LiDAR), a camera, or radio detection and ranging (RADAR))
Implementation Method 2
The processor may obtain a plurality of points corresponding to the external object based on a pulse laser signal reflected from the external object and returned
Data Source
AI summary
The present disclosure relates to a vehicle control apparatus and a method thereof. The present disclosure may include a sensor, such as light detection and ranging (LiDAR), and a processor. The processor may generate a first bounding box corresponding to an external object based on contour points representing a shape of the external object, generate a second bounding box based on aligning first vertices included in the first bounding box around a reference axis, determine whether the first bounding box or the second bounding box indicates that the external object does not interfere with an operation region of the vehicle, based on the contour points, second vertices included in the second bounding box, and dynamics information of the external object.