Results of SKAI Video Analytics Pilot Testing for Azot Company

Control on Hazardous Routes: Project Background

Ensuring safety during the transportation of hazardous materials requires a shift from reactive measures to proactive monitoring of driver conditions. As part of the development of its corporate safety system, Azot KAO initiated a pilot project to implement video analytics systems. As Andrey Semenov, Head of the Information Systems Support Department, notes, the company's fleet consists of over 300 vehicles, and truck tractors involved in the most critical transportations were selected for testing.

Onboard System Architecture and Testing Methodology

To achieve this goal, SKAI deployed a comprehensive hardware and software module. Alexander Kharitonov, a representative of the integrator, details the equipment configuration, which includes:

• DMS (Driver Monitoring System) Camera: aimed at the driver to detect signs of fatigue, distraction, and phone usage.
• ADAS (Advanced Driver Assistance Systems) Camera: monitors the road situation, distance, and risks of direct collision.
• Feedback System: an audio speaker in the cabin and a vibration motor built into the seat for instant notification of violations.

The pilot project lasted two months and was divided into two stages. The first was "silent mode," where the system only collected data on usual driving habits without notifying the drivers. The second was active mode, which included audio and vibration alerts. This approach made it possible to establish an objective baseline for evaluating effectiveness.

Overcoming Sabotage and Correcting Behavior

The implementation of control systems often faces resistance from personnel. During the pilot, speakers recorded cases of intentional sabotage: drivers covered the cameras or turned them away. The solution was the integration of a vibration motor — when attempting to block the view, the system activated physical feedback, which quickly reduced the number of such incidents to a minimum.

During the testing period, the system recorded about 17,000 events. Analysis showed that fine-tuning timeouts and algorithms to real working conditions (for example, increasing the time allowed for assessing the situation in the mirrors) is critical for reducing false alarms and increasing driver trust in the system.

Pilot Results: 18% Reduction in Dangerous Actions

The transition to active mode demonstrated measurable results. Over the two months, the total number of dangerous actions decreased by 18.2%. In particular, cases of driving without a seatbelt, using a smartphone while driving, and signs of drowsiness were reduced. Eliminating these minor but systematic violations directly impacts the reduction of severe accident probabilities, forming a sustainable habit of safe driving.

Project analysts emphasize: technical implementation must be accompanied by explanatory work. Drivers must understand that the system's goal is not to issue fines, but to preserve their lives and health.

What you will learn from this webinar:

• How to properly divide a pilot project into stages to obtain reliable statistics?
• What hardware modules are necessary for comprehensive monitoring of the cabin and road conditions?
• How to overcome driver sabotage using technical means when implementing video analytics?
• Which driving parameters yield the greatest effect when corrected through audio and vibration notifications?
• What is the base cost of equipping one vehicle and what does it depend on?