ADAS Lane Override Control Using Learned Driver Lane Preferences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Advanced Driver Assistance Systems (ADAS) for vehicles lack the ability to autonomously select and maintain an optimal lane of travel based on vehicle-specific and environmental factors, leading to inefficiencies and potential safety issues.
Innovation Solution
A vehicle system that uses a computer with a processor and memory to determine a lane override command, selecting a target lane based on prior travel lane selections, vehicle-specific factors, and environmental data, and actuates vehicle components such as propulsion, braking, and steering to maintain the target lane, utilizing a trained neural network and machine learning algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a lane-control ADAS selects the default travel lane based on roadway geometry, then the system operates with simple lane selection logic, but it fails to account for vehicle-specific factors and environmental conditions leading to suboptimal lane choices
Solution Approach 1:
The system pre-trains a neural network classifier using historical vehicle operating data, driver preferences, and environmental conditions before actual lane selection. This preliminary training enables the system to make adaptive lane selections without real-time complex computations, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The patent replaces traditional rule-based lane selection logic with a machine learning classifier (neural network) that processes vehicle operating data, driver preferences, and environmental conditions. This substitution enables sophisticated adaptability while managing computational complexity through pre-trained models.
2Productivity
If the ADAS system continuously monitors and adjusts lane selections based on real-time vehicle operating data and environmental conditions, then lane selection optimality improves, but computational load and processing time increase
Solution Approach 1:
The neural network classifier is trained offline using historical vehicle operating data, driver preferences, and environmental conditions. This pre-training transfers knowledge to the model, enabling fast real-time inference with minimal computational energy during actual lane selection operations.
Solution Approach 2:
The system incorporates driver responses to lane selection recommendations as feedback to continuously refine the neural network classifier. This feedback mechanism improves lane selection efficiency over time while maintaining low real-time computational requirements through iterative model refinement rather than continuous complex processing.
3Extent of automation
If the system requires driver input to override or confirm lane changes, then safety and driver control are improved, but automation level and response time are reduced
Solution Approach 1:
The system implements partial automation where the neural network classifier provides lane selection recommendations, and the driver can override or confirm. This partial automation approach balances autonomous control with driver oversight, achieving high automation levels while maintaining safety through driver-in-the-loop validation for critical decisions.
Solution Approach 2:
The system monitors driver behavior and preferences, automatically adjusting lane selection strategies based on learned driver patterns. This self-service capability enables the system to anticipate driver intentions, reducing the need for explicit driver inputs while maintaining safety through continuous alignment with driver preferences.
Data Source
AI summary
A vehicle can operate on a roadway for which a default lane is defined. Upon determining that the vehicle is currently operating on the roadway for which the default lane is defined, a lane override command to select a target lane other than the default lane can be determined based on prior lane selections in the vehicle. A vehicle component can be actuated based on the target lane override command.