AI Vehicle Positioning System Using Dynamic Coefficient Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems face inaccuracies in determining the location of vehicles, particularly when traveling through tunnels or remote areas, leading to inefficiencies and potential missed switch points due to loss of location signal or incorrect calculations.
Innovation Solution
A control system utilizing an AI-based model that receives GNSS signals and adjusts coefficients based on errors calculated between GNSS-derived locations and AI-estimated positions, incorporating unscented Kalman filters and fuzzy inference systems to improve accuracy and handle uncertainties, allowing for reliable location determination even without GNSS signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional GNSS-based positioning is used, then the system is simple to implement, but the position determination becomes inaccurate when GNSS signals are lost or in remote locations
Solution Approach 1:
The positioning system is segmented into multiple independent components: GNSS receiver, inertial sensors (accelerometers, gyroscopes), wheel speed sensors, and an AI-based integration unit. Each component operates independently and contributes specific data, allowing the system to maintain accuracy when one component (GNSS) fails while keeping individual components relatively simple
Solution Approach 2:
The patent merges multiple positioning methods (GNSS, inertial navigation, odometry) into a unified AI-based model that processes all inputs simultaneously. This combination allows the system to leverage the strengths of each method while compensating for their weaknesses, achieving high accuracy without requiring any single component to be overly complex
2Productivity
If control systems automatically adjust throttle based on position, then operational efficiency increases, but incorrect position determination causes counterproductive control actions
Solution Approach 1:
The system implements continuous feedback by constantly comparing AI-estimated position with actual GNSS position when available. The AI model uses this feedback to calculate errors and dynamically adjust coefficients, ensuring that automated control actions are based on accurate position information and correcting deviations in real-time
Solution Approach 2:
The AI-based model performs preliminary position estimation using inertial sensors and wheel data before GNSS signal loss occurs, continuously updating the estimate during signal loss. This preliminary action ensures that control systems have reliable position information ready before it's needed, preventing incorrect control actions
3Measurement precision
If multiple variables are used to determine location, then positioning accuracy may improve, but incorrect weight assignment to variables reduces measurement precision
Solution Approach 1:
The system dynamically adjusts the weight (coefficient) assigned to each positioning variable based on current conditions and error calculations. The AI model learns from discrepancies between estimated and actual positions, automatically optimizing the importance of different sensors (GNSS, accelerometers, gyroscopes, wheel data) for different operating scenarios without requiring manual reconfiguration
Data Source
AI summary
A method is provided that may include receiving a global navigation satellite system (GNSS) signal for a moving vehicle and obtaining a position estimation of the vehicle by inputting values of observable characteristics into an artificial intelligence (AI)-based model and receiving the position estimation as an output from the AI-based model. The AI-based model applies different coefficients to the values of the observable characteristics to output the position estimation. The method may also include calculating an error between a location derived from the GNSS signal that may be received and the position estimation obtained from the AI-based model, and changing one or more of the coefficients applied to the observable characteristics in the AI-based model based on the error that is calculated.