Possible with SW alone, without a dedicated board
Precision maps are updated with deep learning.

The Electronics and Telecommunications Research Institute (ETRI) and domestic electric vehicle manufacturer IT Engineering have developed Level 3 and 4 autonomous vehicle technology that allows users to call for a ride via mobile phone, and successfully demonstrated it on July 19.

According to the Society of Automotive Engineers (SAE)'s autonomous vehicle technology levels, Level 3 means the car can perform basic driving tasks, but a driver must take over in hazardous situations. Level 4 means the car can drive itself even in hazardous situations.

IT Engineering's autonomous vehicle, which uses ETRI's technology, automatically comes to the passenger's location when the passenger inputs the departure and destination points into the smartphone and calls out the destination using the voice recognition function, and autonomously drives the passenger to the destination.
In this demonstration, IT Engineering's autonomous vehicle was parked in front of Research Building 3. When an ETRI researcher called it via smartphone, it autonomously drove to Research Building 1, where the researcher was located. The vehicle carrying the researcher autonomously traveled to its destination, Research Building 3, reacting immediately to various unexpected situations. The autonomous vehicle stopped after seeing the signal, pedestrians, and vehicles cutting in, and stopped exactly in front of Research Building 3.
For this demonstration, ETRI equipped the vehicle with two low-cost camera sensors, one lidar sensor, and used one small PC and one standard PC to operate the recognition, judgment, and control software. ETRI explained that the researchers' technology will enable them to directly install components and other components into regular vehicles in the future.

The core technology of this demonstration is one that automatically creates and updates a high-precision map that helps autonomous vehicles perceive their surroundings. This technology allows autonomous vehicles to perceive the surrounding road conditions based on sensor data and a high-precision map, and then precisely updates the high-precision map based on this perception. The margin of error is also small, within 10 cm.

Existing autonomous vehicles require power supplies exceeding hundreds of watts to simultaneously process a vast amount of sensor data and operate autonomous driving AI algorithms. Consequently, development has focused solely on mid- to large-sized sedans and SUVs. However, ETRI has successfully integrated AI algorithms and optimized software to run autonomous driving algorithms using less than 100 watts (W), the equivalent of two laptops.

ETRI is currently conducting research to improve the performance, stabilize, and optimize algorithms to advance autonomous driving technology using deep learning technology. The company announced plans to share data collected so far, including road characteristics and real-time traffic information acquired from cameras, lidar sensors, etc., with universities and companies conducting research on autonomous driving.

The head of ETRI's Advanced Autonomous Driving System Research Group said, "The success of this demonstration has lowered the high barrier to entry for autonomous driving technology development. We expect autonomous driving technology to establish itself as a new convergence industry that generates value while in motion, rather than simply a means of transportation."

IT Engineering CEO Kim Young-han explained, "With the help of our research team, we are witnessing a ray of hope for the mass production of electric vehicles. We will continue to strive to secure future mobility service technologies and enhance market competitiveness."

This technology was developed as a result of the Ministry of Science and ICT's "Development of Collaborative Driving Situation Recognition and Open Platform Technology for Cloud-Based Autonomous Mobility Services." This development led to 26 domestic and international patent applications, 26 academic papers, and 7 technology transfers.

ETRI announced plans to focus on developing autonomous vehicles that can be used by the elderly and physically challenged who cannot drive. To achieve this, the company plans to focus on developing autonomous vehicles that can support an aging society and facilitate transportation in areas with limited public transportation.