Sensor Bias Estimation Using Road Geometry Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor bias estimation techniques in tracking systems rely on cooperative or stationary targets, which are often unavailable, and provide only relative bias estimates, limiting their effectiveness in real tracking situations.
Innovation Solution
A system that estimates sensor bias using known road data within a tracking system, employing road location and sensor observation data to form bias measurements, and updates these estimates over time using Kalman filtering, eliminating the need for cooperative or stationary targets and providing absolute bias estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cooperative or stationary targets are used for sensor bias estimation, then bias estimates can be obtained, but the availability and applicability are severely limited in real tracking situations
Solution Approach 1:
The patent uses known road geometry data as a copy or representation of the actual target positions. Instead of requiring physical cooperative targets, the system creates virtual targets based on mapped road locations, which can then be used for bias estimation in real tracking scenarios without needing actual cooperative targets present
Solution Approach 2:
The patent introduces road location data as an intermediary between the sensor observations and the bias estimation process. This intermediary provides a known reference framework that mediates the comparison between sensor measurements and true target positions, enabling bias estimation without direct access to cooperative targets
2Measurement precision
If multi-sensor comparisons are used for bias estimation, then relative biases can be estimated, but absolute bias estimates and common coverage region requirements are limited
Solution Approach 1:
The patent extracts the reference information from the environmental structure (road geometry) rather than requiring it from multiple sensor comparisons. By taking out the known road locations as an independent reference, the system eliminates the need for multi-sensor overlap regions and can estimate absolute biases without relying on inter-sensor comparisons
Solution Approach 2:
Each sensor can independently perform bias estimation by comparing its own observations against the known road geometry, without needing to coordinate with or compare data from other sensors. This self-service approach allows individual sensors to self-calibrate using the environmental reference, eliminating the complexity of multi-sensor coordination
Data Source
AI summary
A method for measuring sensor bias and a method for estimating sensor bias is provided. Road location data is obtained. Sensor observation data is obtained. A sensor observation is identified that corresponds to an on-road target moving on a known road. A measurement of one-dimensional sensor bias is formed. As one-dimensional sensor bias measurements accumulate, multi-dimensional sensor bias is estimated. Sensor bias estimates may then be incorporated into a multiple hypothesis tracker system.