Hyundai completes development of voice-recognition assistant for cars

A rendering of Hyundai's connected car cockpit to be displayed at the Consumer Electronics Show in Las Vegas in January.
A rendering of Hyundai's connected car cockpit to be displayed at the Consumer Electronics Show in Las Vegas in January.PHOTO: HYUNDAI MOTOR

SEOUL (THE KOREA HERALD/ASIA NEWS NETWORK) - Hyundai Motor has completed the development of a voice-recognition personal assistant jointly developed with a US tech company, and will adopt the service in cars to be released in 2019, the company said on Thursday (Dec 21).

The voice-recognition platform called Intelligent Personal Agent controls in-car systems and features, and provides information for drivers on a real-time basis, the company said in a statement.

The South Korean carmaker will unveil the connected car service at the Consumer Electronics Show (CES) in Las Vegas next month (January 2018). It also plans to demonstrate the system here by deploying the system in its autonomous vehicle powered by fuel cells in February.

The project was jointly developed with SoundHound, a Silicon Valley-based tech firm that specialises in voice-enabled artificial intelligence and conversational intelligence.

Intelligent Personal Agent is designed to interact with SoundHound's conversational intelligence platform Houndify, which responds to drivers' orders and questions through its networks.

The services include making phone calls, receiving and sending text messages, and searching addresses or destinations upon drivers' request.

Searching for music based on the names of artists, album titles and genres is also possible through voice commands, as well as checking the weather and drivers' personal schedules. In-vehicle functions, such as controlling the air conditioning, the operation of sunroofs and door locks, are also possible, the company added.

The system only recognises English for now. The carmaker plans to develop a program that recognises other languages including Korean.

Hyundai will be the first company in the world to demonstrate the system through a connected car cockpit during the CES 2018, officials said.

The system recognises multiple commands and separately processes information when asked more than two questions in one sentence.

The system has also integrated Hyundai's voice user interface and SoundHound's voice-recognition and conversational AI platform. The two companies have been collaborating since 2012. Hyundai Motor Group is the only carmaker that has been investing in SoundHound, officials added.

The US company's technology level was ranked second after China's Baidu in terms of accuracy of voice recognition, according to US analyst Mary Meeker. It was also ranked fourth by the US magazine Fast Company on its list of top 10 AI tech companies worldwide.

Separately, Hyundai Mobis is developing a driverless system that would direct the vehicle to a safe area when it senses the driver dozing off, the company said on Thursday.

The system, Departed Driver Rescue and Exit Maneuver, or DDREM, takes full control of the steering wheel when it determines that the driver is incapable of driving and leads the vehicle to the nearest road shoulder or resting area.

The company aims to fully develop it by 2021, after introducing it at the CES next month.

"Safety is a top priority at Hyundai Mobis, so we are trying to dramatically reduce the number of accidents through DDREM," said David Agnew, the director of advanced engineering at Hyundai Mobis North America, in charge of the development.

According to the American Automobile Association, the number of fatal accidents caused by drowsy driving reached 6,400 cases in 2015.

DDREM is operated with two systems, one of which checks the eye movement of a driver through a biometric sensor and the other alerts unstable lane-changing.

Test runs of the technology began this October at the company's test bed in Detroit. Hyundai Mobis is aiming to test out DDREM on roads next year.