During its recent AI Day 2021 event, VinAI Company under Vingroup - Vietnam’s largest private conglomerate - has announced its three pioneering technologies for smart cars, with huge potential for application and commercialization.
Driver Monitoring System - DMS
Once successful in developing the Face Recognition technology that can identify human faces covered with masks, VinAI has enhanced this feature with a bigger ambition of targeting the automotive industry.
Aimed at improving safety and based on the Face Recognition technology ranked 6th worldwide, VinAI’s DMS uses an IR camera and on-board sensors to recognize driver’s status such as sleepiness, drowsiness or risky behaviours including inattention or using phones. Thus, the system is able to issue timely alerts during the whole journey. Notably, VinAIs DMS can also detect the drivers eye movement behind sunglasses.
VinAI’s DMS is also outstanding thanks to its flexibility as it is designed to be compatible with various hardware platforms from different carmakers, thereby cutting additional production costs. In addition, Euro NCAP (The European New Car Assessment Programme) has already required a driver monitoring system to be one of the primary safety features for any car model looking to receive the sought-after five-star safety rating. By 2022, DMS will become mandatory across the European Union for M1, M2, M3, N1, N2 and N3 new vehicle categories. This shows that DMS is a solution with high commercialization values.
Surround View Monitoring - SVM
With the ambition to bring technological breakthroughs to the automotive industry, VinAI has come up with a solution to universalize surround view monitoring that is often available only for luxury cars. SVM provides 360-degree surround views including top view, rear view, tyre view and 3D view to assist the driver through busy streets, narrow roads or in parking lots.
Using VinAI’s SVM system, drivers can keep an eye on the whole exterior scene outside the car and switch between different view modes using the on-screen control panel. The 3D mode can even allow drivers to observe blind spots that cannot be seen through the mirrors, thereby, detecting obstacles and avoiding unexpected accidents on narrow roads or in parking lots. In addition, SVM well supports the parking process as drivers can fully observe the front, rear, left and right sides.
VinAI’s SVM system is optimized to deliver best-in-class image stitching quality along with high performance and low CPU consumption. The optimization also makes the system flexible to run on different hardware platforms so that it can be integrated to various cars of different segments. In the future, VinAIs goal is to upgrade the SVM system to achieve automatic parking ability by recognizing parking lines, objects such as pedestrians, motorbikes, and obstacles in blind spots, etc. This is an important application that helps to achieve a high-level of autopilot in parking lots or on narrow roads.
Autopilot L2+ features
In recent years, autonomous vehicles have become a trend in the automotive and technology industries to enhance the users driving experience. Solving the autopilot problem requires two key factors: Perception - a module that collects and processes data from the external environment and a Planning & Control module that helps make decisions on the cars actions. These two factors can be understood as the ears, eyes and the brain of the vehicle, which need to work together perfectly to not only ensure safety but also provide comfort when taking over the dynamic driving task.
Regarding Perception, VinAI has successfully developed the lane and 3D object detection features with high accuracy and compatibility to Vietnams conditions, which are still a big problem for preceding solutions in the global market. The algorithm is capable of synthesizing & analyzing data from cameras, radars, maps and vehicle sensors to make optimal control decisions on speed and steering angle for the car.
In addition, VinAIs autopilot features have initially reached level 2+: users do not need to control the steering wheel, gas pedal or brake pedal, but the car can drive by itself safely on the road based on information coming from the perception module. Earlier this year, VinAI successfully completed testing with the lane-keeping Assist and self-driving features on highways and in city roads with low traffic and auto-parking, while focusing on developing the Minimum Risk Maneuver, a stepping stone to reach autopilot level 3 and above.
As shared by Dr. James Tran Duc Trong, leader of Planning & Control team: “VinAI has just approached the autopilot problem for 6 months but has reached level 2+ and at the same time, done with the proof-of-concept. From my point of view, this is a very big step. Keeping up the current speed, VinAI will soon be able to catch up and compete directly with companies in the world that have been many years ahead of us in autonomous vehicles."
With the current level of autopilot, the driver still needs to pay attention and be ready to take control of the vehicle when encountering situations beyond the systems ability to handle. Therefore, the combination with the driver monitoring system will increase the safety of the automated driving feature. The future goal of VinAI is to conquer higher levels of autopilot based on the current foundation, in order to further improve the safety and convenience for users.
After more than 2 years of establishment & development, VinAI has proven its competency in completing research works and commercializing AI applications on a global scale. VinAIs significant products such as face recognition, GuardPro and technologies for smart cars not only demonstrate the breakthrough scientific research works, but also affirm its competitiveness in offering outstanding solutions with high applicability, pioneering in solving challenging problems in the industry and meeting the urgent needs of the society.