Track-Guided Vehicle Localization Using Sensor-Fused Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing track-guided vehicle positioning methods, such as those using odometry and object detection, are prone to errors and require frequent retraining due to environmental changes, compromising safety and reliability in rail traffic operations.
Innovation Solution
A method utilizing a tracking device with non-imaging and imaging sensors to generate location information, employing computer-aided object recognition and machine learning to adaptively create, validate, and utilize data sets for reliable vehicle localization, incorporating redundant tracking methods to ensure continuous operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If object detection methods are used for positioning, then positioning can be achieved without absolute positioning infrastructure, but reliability deteriorates due to environmental changes requiring frequent retraining
Solution Approach 1:
The system performs preliminary training during depot停留 periods when trains are not in regular service. Training data is collected and the object detection model is retrained in advance, so that when regular operations resume, the system already has updated positioning capabilities without interrupting service
Solution Approach 2:
The system uses images captured during normal train operations to automatically collect training data. The object detection system trains itself continuously by processing images it already captures for other purposes, eliminating the need for separate data collection missions or manual intervention
2Reliability
If frequent retraining is performed to maintain reliability, then positioning accuracy is maintained, but train operations are restricted and productivity decreases
Solution Approach 1:
The system performs preliminary training during depot停留 periods when trains are not in regular service. Training data is collected and the object detection model is retrained in advance, so that when regular operations resume, the system already has updated positioning capabilities without interrupting service
Solution Approach 2:
The system implements periodic training cycles that alternate between regular operations and training phases. During depot停留 periods, the system collects data and retrains the model, then returns to normal operations. This periodic approach ensures the model stays current while minimizing disruption to train schedules
3Speed
If conventional odometry is used for speed determination, then speed can be measured continuously, but accuracy deteriorates due to wheel slippage
Solution Approach 1:
The system introduces imaging sensors as an intermediary measurement method. Instead of relying directly on wheel rotation (which slips), the system uses images of track features as an intermediate reference frame to calculate speed, providing an alternative measurement path that is immune to wheel slippage
Solution Approach 2:
The system merges conventional odometry with image-based speed measurement. The odometry provides continuous short-term speed data while image processing provides periodic absolute speed references. By combining these two methods, the system maintains continuous speed measurement capability while correcting accumulated errors from wheel slippage
4Measurement precision
If absolute positioning infrastructure (balises, GPS) is deployed, then positioning accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The system creates a virtual copy of the physical track environment by extracting and storing images of track features (signposts, buildings, terrain). This visual map serves as a digital twin that the object detection system uses for positioning, replacing the need for physical balises or GPS receivers at each location
Solution Approach 2:
The system replaces mechanical positioning infrastructure (balises mounted on tracks, GPS satellites) with an optical system using standard imaging sensors. Instead of mechanical or satellite-based positioning, the system uses computer vision to detect track features and determine position, substituting a simpler, more versatile technological approach
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
The invention encompasses the following: a method for locating a track-guided vehicle on a route, wherein a measurement result is generated with at least one sensor of the locating device, and location information representing the location of the vehicle is calculated from the measurement result, wherein either in a first routine a first sensor, which is not an imaging sensor, is used and the calculated location information is used to train a locating system for the vehicle by means of computer-aided object recognition based on images generated as a measurement result by an imaging second sensor, wherein objects recognized in the images are each stored linked with the location information calculated from the measurement result of the first sensor at the time of image acquisition, and/or in a second routine the first sensor is used and the calculated location information is utilized.to validate vehicle location by object recognition using images generated by the second sensor, wherein objects detected in the images are compared with stored objects linked to stored location information, and if a match is found, the stored location information is compared with the calculated location information and object recognition is validated; and/or in a third routine, the second sensor is used to generate images, objects detected in the images are compared with stored objects linked to stored location information, and if a detected object matches a stored object, the stored location information is used to control (ATP) and/or monitor (ATP) the vehicle movement.where each of the aforementioned first, second, and third routines is executable by the computing environment.