Camera-Based Collision Risk Assessment Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automotive safety systems using lidar and radar sensors often produce false positives due to difficulties in discriminating between types and locations of objects, leading to incorrect collision risk assessments.
Innovation Solution
The implementation of computer vision techniques using machine learning algorithms, specifically convolutional neural networks, to process images from cameras and determine the presence, location, and size of objects, thereby assessing collision risk by identifying objects within a predicted travel path and proximity to the vehicle.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lidar or radar sensors are used for collision risk assessment, then the system can detect objects outside the vehicle, but the system produces false positives due to difficulty in discriminating between types and locations of objects
Solution Approach 1:
The patent replaces lidar and radar sensors with a camera-based computer vision system. The camera captures images that are processed by machine learning algorithms to identify objects, their types, and locations. This substitution eliminates the false positive issue because visual recognition allows precise discrimination between different object types (vehicles, pedestrians, signs) and their locations, while maintaining the ability to detect objects outside the vehicle.
Solution Approach 2:
The patent introduces machine learning algorithms as an intermediary between the camera and the collision risk assessment. The algorithms process the visual data, extract features, and make intelligent decisions about object identification and risk assessment. This intermediary layer enables the system to achieve both high measurement precision for object discrimination and reliable collision risk assessment by combining visual data with contextual understanding.
2Measurement precision
If computer vision techniques with machine learning algorithms are used, then object type and location discrimination accuracy is improved, but the system complexity increases
Solution Approach 1:
The patent segments the collision risk assessment system into distinct functional modules: camera capture, image processing, machine learning algorithms for object identification, feature extraction, and collision risk assessment. This segmentation allows each component to be optimized independently and makes the overall complex system more manageable and maintainable while achieving high measurement precision through specialized algorithms.
Solution Approach 2:
The patent employs a universal camera-based system that can detect multiple types of objects (vehicles, pedestrians, road signs) and perform multiple functions (object identification, location determination, size estimation, collision risk assessment). This multi-functionality reduces the need for multiple specialized sensors and systems, thereby managing complexity while achieving comprehensive measurement precision.
Data Source
AI summary
An image may be obtained from one or more cameras coupled to a first vehicle. The image may be provided as input to a machine learning algorithm configured to determine whether an object depicted in the image corresponds to another vehicle and to determine size information and location information for the object. Output from the machine learning algorithm enables obtaining features including size and location information for a second vehicle that is identified in the image. The features may be used to determine whether the second vehicle is depicted within a predetermined region of the image including a predicted travel path of the first vehicle. The features may also be used to determine whether the second vehicle is within a predetermined proximity of the first vehicle. Thereafter, a determination may be generated as to whether there is a significant risk of collision between the first vehicle and the second vehicle.