Moving Object Position Calculation Using Stored Sensor Landmarks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing position detection methods for moving objects, such as railway vehicles, automobiles, and robots, face challenges in achieving high accuracy and speed without relying on GNSS, especially in environments where GNSS is unavailable, and methods using wheel diameter and reference images are prone to errors and delays.
Innovation Solution
A position information calculation device and method that utilizes sensor information to detect landmarks, associate them with stored position information, and calculate the object's position based on this data, potentially combined with AI learning models and additional position measurement techniques like GNSS, to enhance accuracy and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If wheel diameter and wheel rotation rate are used to detect movement distance, then position detection can be performed without GNSS, but error increases at the time of wheel idling
Solution Approach 1:
The patent introduces an intermediary approach by using image recognition of landmarks as a mediator between wheel-based odometry and true position. The system combines wheel rotation data with visual landmark detection, using the landmark recognition as an intermediary verification mechanism to correct accumulated errors from wheel-based measurement, especially during wheel idling conditions.
Solution Approach 2:
The system dynamically changes measurement parameters by switching between wheel-based odometry and landmark-based visual positioning depending on operating conditions. When wheel idling is detected or accuracy requirements increase, the system transitions to using image-based landmark parameters for position calculation, adapting the measurement approach to current operational parameters.
2Measurement precision
If many reference images are required for position calculation, then position accuracy can be improved, but it becomes difficult to calculate position information of high-speed traveling vehicle without delay
Solution Approach 1:
The patent applies preliminary action by pre-processing and storing landmark information in a database before actual position calculation is needed. Reference images and their corresponding position data are prepared in advance, allowing the system to quickly match current sensor data against pre-processed references without performing intensive computations in real-time, thus maintaining both accuracy and speed for high-velocity applications.
Solution Approach 2:
The system extracts only the essential and most distinctive landmark features from reference images for comparison, rather than processing entire images. By taking out key identifying characteristics of landmarks and storing them in a condensed format, the system reduces computational burden while maintaining position measurement precision, enabling fast processing even when multiple landmarks are involved.
3Measurement precision
If GNSS and autonomous navigation are used in combination to narrow down reference images, then position accuracy can be improved, but a highly accurate position measurement technique is essential
Solution Approach 1:
The patent segments the position measurement task into distinct functional modules: GNSS provides coarse position information, autonomous navigation refines this using landmark recognition, and each module operates independently with clearly defined input-output interfaces. This segmentation allows the system to achieve high precision through coordinated simple techniques rather than requiring a single complex highly accurate measurement technique, reducing overall system complexity while maintaining precision.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A position information calculation device (20) of an embodiment includes a sensor information acquisition module (20A), a landmark detection module (20B), a landmark dictionary section (20D), a landmark search module (20C), and a moving object position calculation module (20F). The sensor information acquisition module (20A) is configured to acquire sensor information obtained by sensing an external environment of a moving object (RV) with a sensor. The landmark detection module (20B) is configured to detect a predetermined landmark set in advance from the sensor information to acquire first sensor landmark information. The landmark dictionary section (20D) is configured to store in advance second sensor landmark information in which the landmark and position information are associated with each other. The landmark search module (20C) is configured to acquire the position information associated with the landmark in the first sensor landmark information by search using the landmark dictionary section. The moving object position calculation module (20F) is configured to calculate a position of the moving object (RV) on the basis of the position information of the landmark.