Autonomous Vehicle Lighting State Determination via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inaccurate determination of vehicle lighting states can lead to collisions due to ambiguity and insufficient data, affecting the behavior prediction and trajectory generation for autonomous vehicles.
Innovation Solution
An autonomous vehicle employs a machine learning algorithm, such as a convolutional neural network, to analyze image data from sensors and determine probabilities of various lighting states (e.g., turn signals, brake lights) of tracked objects, enabling accurate trajectory planning and prediction of their behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional lighting state determination methods are used, then the system is simple, but the accuracy is insufficient leading to collisions
Solution Approach 1:
The patent replaces traditional rule-based lighting state determination with a machine learning model that processes sensor data. The model uses image data from cameras and point cloud data from LiDAR to classify lighting states (brake light, turn signal, hazard light, or none), achieving higher accuracy without mechanical complexity increases.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between raw sensor data and lighting state determination. This intermediary processes multiple data sources (image data, point cloud data) and temporal information to resolve ambiguities that traditional direct methods cannot handle.
2Measurement precision
If more sensor data is collected to resolve ambiguity, then the accuracy improves, but the data processing complexity increases
Solution Approach 1:
The patent merges multiple data sources including image data from cameras, point cloud data from LiDAR, and temporal information from sequential frames. The machine learning model processes these combined data streams together to determine lighting states, resolving ambiguities that single data sources cannot address.
Solution Approach 2:
The patent performs preliminary processing of sensor data by extracting relevant features (image crops, point cloud subsets) before feeding them to the machine learning model. This preliminary action reduces the complexity of the main classification task while preserving the information needed for accurate lighting state determination.
3Speed
If real-time lighting state determination is implemented, then the response time is fast, but the accuracy may be insufficient due to ambiguous data
Solution Approach 1:
The patent uses periodic sampling of sensor data at defined frame rates to achieve real-time processing. The machine learning model processes data at regular intervals, balancing the need for fast response with sufficient data collection for accurate classification of lighting states.
Data Source
AI summary
Techniques for determining lighting states of a tracked object, such as a vehicle, are discussed herein. An autonomous vehicle can include an image sensor to capture image data of an environment. Objects such can be identified in the image data as objects to be tracked. Frames of the image data representing the tracked object can be selected and input to a machine learning algorithm (e.g., a convolutional neural network, a recurrent neural network, etc.) that is trained to determine probabilities associated with one or more lighting states of the tracked object. Such lighting states include, but are not limited to, a blinker state(s), a brake state, a hazard state, etc. Based at least in part on the one or more probabilities associated with the one or more lighting states, the autonomous vehicle can determine a trajectory for the autonomous vehicle and/or can determine a predicted trajectory for the tracked object.