Vehicle Light Indicator Label Correction for Autonomous Driving AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems face challenges in accurately detecting light indicators on vehicles, leading to incorrect predictions that can affect vehicle control, due to insufficient or inaccurate labeling of images used for training machine learning models.
Innovation Solution
A system and method for labeling images by obtaining vehicle images, identifying vehicle positions, displaying graphical indicia, and receiving user inputs to correct light indicator status, specifically focusing on brake lights and turn signals, to improve the accuracy of machine learning models used in autonomous driving systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If images are obtained from autonomous driving systems for training machine learning models, then the quantity of training data is increased, but the accuracy of light indicator detection remains insufficient due to improper determination by the autonomous driving system
Solution Approach 1:
The system implements a feedback mechanism where autonomous driving systems send images with improper light indicator determinations to a server. The server then provides corrected labels back to the autonomous driving systems, creating a closed-loop feedback system that continuously improves detection accuracy while utilizing the existing data infrastructure.
Solution Approach 2:
The autonomous driving systems automatically identify and flag their own improper determinations, sending these self-identified problematic images to the server for correction. This self-service approach allows the system to automatically generate training data from its own errors without requiring external intervention for data collection.
2Adaptability or versatility
If machine learning models are trained with large quantities of images, then the model's generalization capability is improved, but the reliability of light indicator detection is reduced due to inaccurate labeling
Solution Approach 1:
The server acts as an intermediary between autonomous driving systems, receiving improperly determined images and providing corrected labels. This intermediary ensures that training data is both diverse (maintaining generalization) and accurately labeled (maintaining reliability) by centralizing the correction process.
Solution Approach 2:
The system performs preliminary correction of light indicator labels before the images are used for training the machine learning model. By pre-correcting the labels on the server side, the model receives high-quality training data that maintains both diversity for generalization and accuracy for reliability.
3Ease of operation
If autonomous driving systems detect light indicators on vehicles, then vehicle control decisions are enabled, but false predictions occur leading to incorrect control actions
Solution Approach 1:
The system performs preliminary correction of light indicator detection errors before the autonomous driving system makes control decisions. By pre-training the machine learning model with corrected labels, the system reduces false predictions and improves the reliability of subsequent control actions.
Solution Approach 2:
The system implements feedback loops where control decisions and their outcomes are monitored. When improper determinations are detected, the system sends images to the server for correction and uses this feedback to continuously improve the accuracy of future control decisions.
Data Source
AI summary
This application relates to systems and methods to train a machine learning model used for autonomous driving. The system includes a plurality of vehicles configured to capture at least the front surrounding view of the vehicle, a machine learning training system, and a verification computing device. The machine learning training system is configured to receive the captured images from the vehicles. The verification computing device is configured to verify whether the machine learning model correctly identified the light indicator of vehicles shown in the captured image. The verification device may determine a disagreement between the vehicle's predicted light indicator and the correct light indicator. In determining that at least one vehicle has a disagreement, the verification computing device is configured to modify the light indicator label and correct label. Then, the modified label can be fed into the machine learning model and used for training the machine learning model.