Hyundai Motor Group has developed what it says it the world’s first Machine-Learning-based Smart Cruise Control (SCC-ML)—a technology that incorporates the driver’s patterns into its self-driving behavior, creating a custom experience for the driver.

The technology incorporates artificial intelligence (AI) within the Advanced Driver Assistance System (ADAS) feature. The system is planned for implementation in future Hyundai Motor Group vehicles.

The new SCC-ML improves upon the intelligence of the previous ADAS technology to dramatically improve the practicality of semi-autonomous features. Hyundai Motor Group will continue the development efforts on innovative AI technologies to lead the industry in the field of autonomous driving.

—Woongjun Jang, VP at Hyundai Motor Group

Smart Cruise Control (SCC) enables an essential self-driving feature and core technology for ADAS: maintaining distance from the vehicle ahead while traveling at the speed selected by the driver.

SCC-ML combines AI and SCC into a system that learns the driver’s patterns and habits on its own. Through machine learning, Smart Cruise Control autonomously drives in an identical pattern as that of the driver.

To operate earlier versions of Smart Cruise Control technology, the driver had to manually adjust driving patterns—such as the distance from the preceding vehicle and acceleration. Hyundai concluded that it was impossible to fine-tune the settings to accommodate individual preferences without machine learning technology.

For example, the same driver may accelerate differently in high-speed, mid-speed and low-speed environments depending on circumstance—but detailed fine-tuning of SCC was not available. Therefore, when Smart Cruise Control was activated and the vehicle operated differently than desired, drivers sensed the difference, resulting in a reluctance to use the technology because it made them feel anxious and unstable.

Hyundai Motor Group’s independently developed SCC-ML operates as follows: First, sensors, such as the front camera and radar, constantly acquire driving information and send it to the centralized computer. The computer then extracts relevant details from the gathered information to identify the driver’s patterns. A machine learning algorithm is applied during this process.

The driving pattern can be categorized into three parts: distance from preceding vehicles; acceleration (how quickly it accelerates); and responsiveness (how quickly it responds to driving conditions). In addition, driving conditions and speeds are considered as well.

A sample use case is maintaining a short distance from the preceding vehicle during slow, city driving, and then further away when driving in the fast lane. Considering these various conditions, SCC-ML makes analysis to distinguish over 10 thousand patterns, developing a flexible Smart Cruise Control technology that can adapt to any driver’s patterns.

The driving pattern information is regularly updated with sensors, reflecting the driver’s latest driving style. In addition, SCC-ML is programmed specifically to avoid learning unsafe driving patterns, increasing its reliability and safety.

With upcoming Highway Driving Assist system that features automatic lane change assist, SCC-ML achieves SAE Level 2.5 self-driving, Hyundai says.