Innopolis University: A system for assessing a driving test

The main articles are:

• Video analytics (terms, applications, technologies)
• What is the Internet of Things (IoT)

2023: Patent for a system for evaluating a driving test

On April 7, 2023, representatives of Innopolis University announced that they had received a patent for a system for evaluating a driving test. The invention of the employees of the Center for Unmanned Technologies allows you to remotely monitor car control. Thanks to artificial intelligence methods and a control unit of its own design, the device automatically detects 21 types of traffic violations. The device will be able to use driving schools while teaching drivers and passing exams in an urban environment.

System for evaluating a driving test

As reported, the driving control device consists of a device with sensors object recognition installed in the car and on its roof, and remote servers storages maps, routes and exam protocols. When the video camera recognizes a student who has got into the car, the system allows you to start moving and the inspector starts the program in the application on. tablet During the exam, video and sound are recorded inside and outside the vehicle.

The device automatically records speeding, entering the oncoming lane, non-allowing pedestrians on the zebra, non-compliance with the requirements of priority signs and traffic lights. Errors related to the human factor, for example, driving along the roadway at a low speed unnecessarily, uncertain driving, incorrect interpretation of the instructions of the controller, the inspector notes in the tablet. Data is sent to a remote server during or after the exam, and a report is downloaded there.

During the exam or in vehicle control mode, the neural network acceleration module processes information using artificial intelligence techniques. Thanks to cameras and radars, the system fixes a pedestrian on a zebra, while predicting the trajectory of his movement. If a person changes the original direction, or stops in front of the car, then the device will record a violation "did not yield to the pedestrian." A similar mechanism is used when passing intersections and skipping vehicles according to rules and priority signs. told Evgeny Fadeev, Deputy Director of the Center for Unmanned Technologies of Innopolis University

Algorithms for recognizing and predicting motion paths are launched near intersections and pedestrian crossings. Alternately turning on and off individual algorithms optimizes computing power and reduces the cost of equipment by 10-15%. This feature of the system also allows you to use a smaller computer, so that the device can be easily integrated instead of a radio under the car panel.

System for evaluating a driving test

The customer of the project - the State traffic inspectorate of the Ministry of Internal Affairs of the Russian Federation. Prototypes of the device were tested on the roads of Innopolis and Kazan. According to the developers, for the correct operation of the system, you need pre-prepared digital maps of the exam area, several tests for training object recognition systems, and on the day of the exam - weather with good visibility. If the development of Innopolis University is introduced into the work of the interdistrict registration and examination departments of the traffic police, driving schools will also use the device.

Development will make the exam safer, and grading more transparent and objective. At the same time, it will reduce the load on the inspector: due to the automation of many processes, he will perform fewer actions during the exam. For example, the system independently builds a route for the exam and detects errors. In addition, most of the inspector's decisions will be rechecked by the device, that is, to act as a "second opinion" and an assistant in controversial situations during the exam. told Evgeny Fadeev, Deputy Director of the Center for Unmanned Technologies of Innopolis University

Detailed description and features of the invention are published by the Federal Service for Intellectual Property of the Russian Federation. The "device and method for controlling the control of a vehicle" are patented with the assistance of the Intersectoral Technology Transfer Center of Innopolis University.

The development combines several technological solutions for vehicle control at once: sensors and sensors with intelligent systems for detecting and recognizing road situations, systems for processing data on the location of a car. The patented device can be used not only when conducting an exam for a driver's license. Individual algorithms can be integrated into driving simulators for classroom training for driving school students. All these factors make the product unique and increase its potential for commercialization. reported by Marina Abdrakhmanova, head of the Intersectoral Technology Transfer Center of Innopolis University