In a normal driving test, the driving examiner sitting next to the driving candidate is required to monitor the driver’s actions as well to keep track of the peripherals. To put forth the safety as the driving examiner prime focus, it is wished to leverage on AI technologies to assist driving examiners to assess the driving performance of the driver candidates, along with that more accurate and objective driving information could be obtained by sensors and cameras installed on the vehicles and surroundings.
As a first trial to start with, the driving test will only consist of test items in the basic test and be conducted within the driving school campus. Vision at high position to overview the area of interest is preferred.
Project Challenges:
Different angle, lighting condition, camera quality and occlusion shall be considered;
Installation work should be non-intrusive subject to available installation locations; power supply and data connectivity shall be assessed in advance;
Test cases involved in the basic test should be referenced to the normal driving test for the public, which includes but not limited to, “S”/”L” parking, turning around on a narrow road, starting or stopping on a slope, etc.
Expected Outcome
An AI-assisted system for analysing the whole process of a basic driving test and generate the assessment report;
Video recording of each entire whole driving test process including the corresponding AI analytical data;
Trained AI model(s) that can provide correct assessment for a series of car motions and driving behaviors concerned in the basic test;
An overall of 90% or above accuracy is preferred;
All hardware required for the system including provision and installation, e.g. cameras, sensors, AI processing server, data storage, etc.
Expected Trial Duration
9-month
Contact Information
I&T Wish Proposer
:
Electrical and Mechanical Services Department (EMSD)
