Intelligent video analytics for automobile factories
A climate laboratory was built at a factory in China to test new car models. The requirement was not simply to equip the laboratory with cameras, but to create an intelligent system.
Project background
A climate laboratory was built at a factory in China to test new car models — a complex that simulates extreme operating conditions: temperatures from −35°C to +80°C, solar radiation intensity up to 2000 W/m², and vibrational loads. The requirement was not simply to equip the laboratory with cameras, but to create an intelligent system that analyzes vehicle behavior in real time, records assembly and material defects, monitors personnel safety, and integrates with the laboratory's industrial automation.\
Key project challenges:
- personnel safety control under climate test conditions (slippery surfaces, vibration stands, extreme temperatures);
- vehicle testing on a vibration stand — analysis of body vibrations, diagnostics of discrepancies with the stand frequency, and control of wheel position;
- material degradation assessment — change in paint color, melting of plastic parts under the impact of the solar panel and temperature;
- confidentiality control — prevention of photo and video recording of new models during the testing phase;
- deep integration with Siemens industrial controllers to synchronize analytics with the laboratory's operating modes.\
Implementation process
Specialists from OSMOVIEW LLC developed and implemented a comprehensive solution for the climate laboratory based on the OSMOCV platform. The platform was deployed on the customer's server infrastructure in a local network.
Cameras were installed in the laboratory (industrial cameras, ultra-high-resolution cameras, high-speed cameras up to 250 FPS;* IP cameras placed at different levels of the laboratory and in the operator's room). Video streams are fed into the OSMOCV platform for real-time processing.
The solution included neural services developed in-house by OSMOVIEW LLC
Personnel safety:
- Fall detection — recording of people falling or stumbling on slippery surfaces and vibration stands, taking into account the multi-level structure of the laboratory;
- Face recognition — identification of employees authorized to work in the laboratory and restricted areas;
- Fire and smoke detection — early detection of fires using standard cameras without the use of thermal imagers;
- Mobile phone usage control — prevention of photo and video recording in the new model testing zone;
- Operator presence monitoring — tracking the time of actual work/absence at the workplace;
- Intruder detection — access control to isolated zones.
Engineering diagnostics (unique scenarios):
- Vibration analytics — using industrial cameras (including ultra-high-speed up to 250 FPS), the system recognizes the first body vibration when the car enters the stand, records discrepancies between the vibration table frequency and the actual body/wheel vibration;
- Paint degradation assessment — detection of paint color changes during temperature and light tests;
- Plastic part degradation assessment — based on changes in the solar panel color and the appearance of the plastic, the system detects part melting, records control points, and provides a quantitative assessment of deformation;
- Wheel position control — detection of wheels on the vibration table stands.\
Integration and management:
- Full integration with Siemens controllers and other laboratory automation devices — the system receives automation status, vibration table operation data, temperature, and humidity;
- Dynamic camera configuration management (e.g., when the solar panel is turned on);
- Periphery management — camera cooling fans and their power supply;
- Scenario validation with automation parameters — analytics checks a set of industrial logic conditions and metrics;
- Generation of email and SMS notifications about critical events.\
Project features and complexities
The project was the first on the OSMOCV platform with such a volume and depth of integration. A wide range of diverse tasks (security, privacy, quality, engineering diagnostics) required a modular approach, where each logic is verified by a set of conditions from the automation system. The customer's high standards for interfaces and documentation required continuous UI/UX refinement and close collaboration with the factory at all review stages. Engineering integration with Mindvision and Siemens equipment was conducted in dedicated sessions with stand tests and extended logging for accelerated debugging. Each scenario was developed and configured individually, as most of the functions in the platform did not yet exist at the start of the project.\
Results
- The OSMOCV platform has been launched for industrial operation in the operator laboratory, with users and roles implemented for operators.
- Full personnel safety is ensured under extreme climate test conditions — automatic detection of falls, fire, intruders, and privacy violations.
- Unique engineering scenarios have been implemented — vibration detection, assessment of paint and plastic part degradation, and wheel position control on a vibration stand.
- Deep integration with Siemens automation has been implemented — analytics are synchronized with laboratory operating modes, temperature, humidity, and the state of the vibration table.
- A scalable foundation has been created for further implementation of AI analytics on other production lines.
- System operability has been confirmed in a wide range of conditions: from −35°C to +80°C, with light intensity up to 2000 W/m².