Traffic Light Lane-State Recognition for Unmapped Autonomous Driving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in detecting and responding to traffic lights without pre-existing map information, leading to potential requests for remote assistance in unmapped or outdated areas.
Innovation Solution
A machine learning model is trained using labeled camera images and map data to identify traffic light locations and lane relationships, enabling the vehicle to determine lane states and respond accordingly, even in unmapped areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous vehicles use pre-stored map information for navigation, then navigation reliability is improved in known areas, but the system fails in unmapped or outdated areas requiring remote assistance
Solution Approach 1:
The system performs preliminary actions by training a machine learning model offline using labeled image data and map information before deployment. The model is pre-trained to recognize traffic lights and determine lane states, enabling the vehicle to independently handle unmapped areas without requiring remote assistance during operation
Solution Approach 2:
The autonomous vehicle uses self-service by employing the trained machine learning model to independently detect traffic lights and determine lane states in real-time. The system serves itself by processing camera images through the model to obtain lane state information without external intervention, eliminating the need for remote assistance in unmapped areas
2Measurement precision
If the system requests remote assistance for unmapped areas, then navigation accuracy is maintained, but operational efficiency and response time deteriorate
Solution Approach 1:
The system achieves self-service by using the trained machine learning model to autonomously detect traffic lights and determine lane states from camera images in real-time. This eliminates the need to request remote assistance, maintaining detection accuracy while significantly reducing response time by providing immediate local processing results
3Device complexity
If the vehicle relies on pre-stored map information, then system complexity is reduced, but the system becomes obsolete in areas without current map data
Solution Approach 1:
The trained machine learning model serves as an intermediary between the simple pre-stored map information and the complex task of traffic light detection and lane state determination. The model bridges this gap by processing camera images to extract lane state information, maintaining low system complexity while enabling adaptability to unmapped areas through image-based recognition
Data Source
AI summary
Methods and system are provided for training and using a model to determine states of lanes of interest. For instance, image data including an image and an associated label identifying at least one traffic light, a state of the at least one traffic light, and a lane controlled by the at least one traffic light are received and used to train the model such that the model is configured to, in response to receiving an image and a lane of interest included in the image, output a lane state for the lane of interest. This model is then used by a vehicle in order to determine a state of a lane of interest. This state is then used to control the vehicle in an autonomous driving mode based on the state of the lane of interest.