Monocular Camera Vehicle Distance Control Using Annotated ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems cannot maintain a suitable vehicle-to-vehicle distance with preceding vehicles due to lack of annotation information regarding distance in training data, leading to inadequate recognition and control.
Innovation Solution
A vehicle control device using a machine learning model generated with teacher data images including annotation information about suitable, shorter, and longer vehicle-to-vehicle distances, allowing the processor to estimate and adjust the distance autonomously based on images from a monocular camera.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If annotation information about vehicle-to-vehicle distance is added to training data, then the recognizer can output distance estimation results, but the complexity of data preparation increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and annotating vehicle-to-vehicle distance information during the data preparation phase. The annotation information including distance categories (shorter than suitable distance, suitable distance, longer than suitable distance) is prepared in advance alongside the image data, enabling the recognition model to learn distance estimation without requiring complex real-time calculations during actual operation.
2Device complexity
If a monocular camera is used instead of multiple sensors, then the device complexity is reduced, but the ability to accurately measure distance is worsened
Solution Approach 1:
The patent replaces the mechanical/optical distance measurement system (multiple sensors) with a machine learning-based recognition system. The monocular camera captures images, and the recognition model processes these images to estimate vehicle-to-vehicle distance by learning from annotated training data, substituting physical measurement mechanisms with computational intelligence.
Solution Approach 2:
The patent changes the parameter representation by transforming continuous distance values into categorical annotations (shorter than suitable distance, suitable distance, longer than suitable distance). This parameter transformation enables the recognition model to effectively learn distance estimation from image data alone, achieving accurate distance measurement without requiring multiple sensors.
3Ease of manufacture
If the recognizer is trained without distance annotation information, then the training process is simpler, but the autonomous driving control cannot maintain suitable vehicle-to-vehicle distance
Solution Approach 1:
The patent applies preliminary action by preparing annotation information about vehicle-to-vehicle distance in advance during data preparation. The training data includes categories indicating whether the distance is shorter than suitable, suitable, or longer than suitable distance, enabling the recognition model to learn accurate distance estimation before deployment, ensuring reliable autonomous driving control.
Data Source
AI summary
A vehicle control device having a vehicle control part and a vehicle-to-vehicle distance estimation part estimating a vehicle-to-vehicle distance between a host vehicle and a preceding vehicle using a machine learning model based on an image including the preceding vehicle captured by a monocular camera mounted on the host vehicle, wherein the machine learning model is generated by performing machine learning using teacher data images including a preceding vehicle for learning captured from a vehicle for capturing teacher data images and annotation information added to the teacher data images, and the annotation information includes information showing any of the vehicle-to-vehicle distance between the vehicle for capturing teacher data images and the preceding vehicle for learning being suitable, the vehicle-to-vehicle distance being short, and the vehicle-to-vehicle distance being long.