Virtual Sensor Data Generation for Railway Crossing Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Developing and testing algorithms for safely navigating railroad crossings and similar obstacles in varying weather conditions is challenging due to the complexity of real-world environments and the need for diverse and unbiased training data.
Innovation Solution
A system that generates virtual sensor data using a combination of computer hardware and software, including vehicle-motion models, sensor models, and simulation modules, to create a virtual driving environment with various weather conditions and obstacles, allowing for the production of diverse and unbiased training data for algorithm development and testing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If real-world sensor data is collected for algorithm training, then the data reflects actual environmental complexity, but the data lacks diversity and coverage of rare scenarios
Solution Approach 1:
The patent creates virtual copies of real-world sensor data by generating synthetic sensor readings from virtual models of railroad crossings, cattle guards, and weather conditions. This allows diverse training scenarios to be generated without physical data collection, resolving the contradiction between data diversity and practical collection limitations
Solution Approach 2:
The system pre-generates comprehensive training data covering rare and extreme weather conditions before algorithm deployment. By preparing diverse scenarios in advance through virtual simulation, the system ensures algorithms are trained on all possible conditions without needing to collect data during actual operation
2Reliability
If algorithms are trained on limited real-world data, then training is faster and easier, but the algorithms fail to generalize to varying weather conditions
Solution Approach 1:
The patent introduces a virtual environment as an intermediary between simple data collection and complex real-world testing. This intermediate virtual layer allows comprehensive algorithm training with controlled complexity, bridging the gap between limited real data and the need for weather-generalization
Solution Approach 2:
The system varies weather parameters (precipitation, temperature, visibility) in the virtual environment to generate diverse training scenarios. By systematically changing these parameters, the system achieves high algorithm reliability across conditions without requiring equally complex physical testing setups
3Quantity of substance
If diverse training scenarios are created through virtual simulation, then data coverage is improved, but the computational resources and system complexity increase
Solution Approach 1:
The patent segments the training system into modular virtual components: virtual sensors, virtual obstacles (railroad crossings, cattle guards), and virtual weather conditions. This segmentation allows selective generation and combination of data elements, achieving diversity without requiring a completely complex simulation system for each scenario
Data Source
AI summary
A method for generating training data is disclosed. The method may include executing a simulation process. The simulation process may include traversing a virtual, forward-looking sensor over a virtual road surface defining at least one virtual railroad crossing. During the traversing, the virtual sensor may be moved with respect to the virtual road surface as dictated by a vehicle-motion model modeling motion of a vehicle driving on the virtual road surface while carrying the virtual sensor. Virtual sensor data characterizing the virtual road surface may be recorded. The virtual sensor data may correspond to what a real sensor would have output had it sensed the road surface in the real world.