Mobile Object Positioning Using Reliability-Weighted Environmental Features
Solution Overview
Problem
Current methods for determining the position and planning paths for moving objects, such as autonomous vehicles, face inaccuracies due to limited angular resolution, reflective surfaces, transparent obstacles, and changing environmental conditions, which are not adequately addressed by existing occupancy probability methods.
Innovation Solution
An environment representation device that stores location-resolved reliability data and occupancy probability data, allowing for improved position determination and path planning by weighting features based on their detectability and presence probabilities, with the ability to update reliability data dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If occupancy probability methods are used to represent environmental features, then the system can handle uncertain detections, but the position determination accuracy deteriorates because reliability variations of different features are not accounted for
Solution Approach 1:
The patent applies local quality by associating reliability values with individual environmental features rather than using a uniform occupancy probability for all features. Each feature (e.g., wall, door, shelf) is assigned a specific reliability value reflecting its detectability characteristics, allowing the system to differentiate between highly reliable and less reliable features in position determination calculations
2Measurement precision
If laser scanning is used to detect environmental features, then the system can obtain precise distance and direction data, but measurement accuracy deteriorates due to reflective surfaces, transparent obstacles, and limited angular resolution
Solution Approach 1:
The patent converts the harmful effects of reflective and transparent surfaces into beneficial information by assigning lower reliability values to features detected under these conditions. Instead of discarding such detections, the system incorporates them with appropriate reliability weighting, allowing partially detectable features to contribute to position determination without compromising overall accuracy
Solution Approach 2:
The patent changes the parameter of feature representation from binary (detected/not detected) to continuous (reliability value between 0 and 1). This allows the system to capture the degree of detection confidence and account for varying detectability conditions, transforming measurement uncertainties into quantifiable reliability parameters
3Reliability
If odometry or stereoscopic image analysis is used for position estimation, then the system can calculate position without external references, but position accuracy deteriorates over time due to cumulative errors
Solution Approach 1:
The patent implements feedback by continuously comparing the vehicle's odometry-based position estimate with the map-based position determined using reliability-weighted feature detections. This feedback mechanism allows the system to correct cumulative odometry errors by referencing the more stable map-based position, maintaining long-term position accuracy
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for determining the position of a moving object in an environment containing electronically detectable features comprises the steps of providing a digital representation of the environment, which represents several electronically detectable environmental features and also specifies or allows the derivation of respective reliability values for at least some of these features; measuring the environment from the perspective of the moving object, wherein the measurement involves capturing several of the electronically detectable environmental features; evaluating the measurement result with data derived from the environment representation, taking into account existing reliability values; and determining the position based on the evaluation result.