Adaptive Steering Geometry Using Reinforcement Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vehicle control systems fail to dynamically adjust steering geometry based on driver behavior, driving conditions, and vehicle application, leading to suboptimal performance and compliance with legal regulations.

Innovation Solution

Implementing a reinforcement learning model and neural network system that utilizes vehicle sensors to determine the current driving cycle and application, and adjusts wheel alignment using actuators or provides recommendations for manual adjustment based on machine learning outputs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If steering geometry is fixed for all conditions, then device complexity is reduced, but vehicle performance and compliance with regulations deteriorate

Engineering Contradiction:
Improvesteering control system complexityVSAvoidvehicle performance compliance
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent implements dynamic steering geometry adjustment by transitioning from fixed alignment to actively adjustable camber and toe settings. The system uses actuators controlled by machine learning models to continuously adapt steering geometry parameters based on real-time driving conditions, vehicle speed, and driver behavior patterns, thereby maintaining optimal performance across varying operational contexts.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes physical parameters of the steering system by adjusting camber and toe angles through actuated mechanisms. The machine learning models determine optimal parameter values dynamically, and the physical steering components are reconfigured to match these parameters, enabling the vehicle to adapt its steering characteristics to different driving scenarios and regulatory requirements.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If steering geometry is dynamically adjusted based on driving conditions, then vehicle performance is improved, but device complexity increases

Engineering Contradiction:
Improvevehicle performance optimizationVSAvoidsteering control system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system implements self-service through autonomous machine learning models that automatically analyze driving conditions and adjust steering geometry without driver intervention. The reinforcement learning component continuously learns from driving data and feedback, enabling the system to self-optimize performance while reducing the complexity of manual adjustment mechanisms.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent incorporates feedback loops where sensor data from vehicle operation is fed into machine learning models that adjust steering geometry parameters. The system monitors performance metrics and regulatory compliance, using this feedback to refine adjustments in real-time, thereby optimizing performance while managing system complexity through intelligent control algorithms.

Inventive Principle:
Principle #23Feedback

3Manufacturing precision

If machine learning models are used to determine wheel alignment, then manufacturing precision is improved, but ease of manufacture deteriorates

Engineering Contradiction:
Improvewheel alignment precisionVSAvoidsystem implementation difficulty
Core Design Contradiction:
Manufacturing precisionVSEase of manufacture

Solution Approach 1:

The patent replaces traditional mechanical alignment systems with machine learning-based electronic control. Instead of relying on fixed mechanical linkages and manual adjustment mechanisms, the system uses sensors, processors, and actuators controlled by trained machine learning models to achieve precise wheel alignment dynamically, thereby improving precision while accepting increased manufacturing complexity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4263327B1System and method for reinforcement learning of steering geometry
Publication Date: 2025.10.08 VOLVO TRUCK CORP
  • EP4263327B1 patent drawingFigure 1A~1B
  • EP4263327B1 patent drawingFigure 2~3
  • EP4263327B1 patent drawingFigure 4

AI summary

Systems, methods, and computer-readable storage media for adjusting the steering geometry of a vehicle by using reinforcement learning in series with a neural network to determine when and how to adjust the steering geometry of the vehicle. A system can do this by receiving vehicle information associated with ongoing movement of the vehicle, and executing a reinforcement learning model using that vehicle information. The outputs of the reinforcement learning model can include a current driving cycle of the vehicle and a current application of the vehicle. The system then executes a machine learning model, where inputs to the machine learning model can include the outputs of the reinforcement learning model and the vehicle information. The outputs of the machine learning model can then include a wheel alignment signal.