By the end of 2029, Hyundai Motor Group will debut vehicles equipped with its own software for driver assistance, representing a significant milestone for the South Korean auto company in achieving more control over autonomous driving technology. The company has delayed the original timing by around two years, and in the interim is using Nvidia technology to accelerate the deployment of more sophisticated driver-assistance technologies.
Hyundai will be collaborating with Nvidia to deliver Level 2+ and Level 2++ driver-assistance features to its cars in 2028 before they roll out their own system. The collaboration shows how artificial intelligence, high performance computing and large-scale driving data is becoming increasingly vital in the automotive industry, where vehicle manufacturers are racing to create increasingly capable technology to take on parts of the driving role.
To be sure, Hyundai’s decision to permit the system to be developed in-house shows just how challenging it is to build complex automated-driving software at scale. However, Hyundai had earlier planned to release the technology in the late 2027. The updated timeline may add to the company’s challenges as Tesla and several Chinese automakers are aggressively pushing into the realm of advanced driver-assistance and autonomous driving technology.
But Hyundai executives have repeated that the deal with Nvidia is not an end to the plans to develop its own technology. Hyundai Motor Group President Park Min-woo explained that the collaboration will be used to speed up the development process, while at the same time allowing Hyundai to gain knowledge and learn useful driving data.
Park Min-woo, a Hyundai Motor Group president, told a media briefing that “our partnership with Nvidia is not all-or-nothing.
Hyundai will collaborate on the technology and leverage data from the systems to train and enhance its own software platform, called Atria, Park said. The way it works provides Hyundai with Nvidia’s computing and car tech, and keeps the automaker developing an independent software platform for future cars.
The strategy marks a turnaround for Hyundai’s software initiatives. Previously serving at Nvidia and coming on board with Hyundai in January, Park has been at the forefront of working to increase collaboration between the two. Not only does his strategy vary from his predecessor Song Chang-hyeon’s, who had been focusing on internal development of autonomous-driving software for more time, but it also differs from the other major players in the industry.
Hyundai’s next generation safety technologies are expected to feature a mix of cameras, radar and ultrasonic sensors, rather than the more costly lidar devices. This sensor combination is likely to be deployed in the first phase of the Nvidia’s Hyperion 10 platform-based vehicles.
The restriction in the use of lidar may help to bring down the costs of advanced driver assistance technology for Hyundai. While cameras and radar are already commonplace in modern cars, lidar systems can be quite expensive in terms of hardware. Meanwhile, sensor selection is still a key technical challenge as autonomous-driving systems need to be able to identify vehicles, pedestrians, road marking, and other objects, in a variety of conditions.
Hyundai isn’t ruling out lidar completely. The company is looking to apply the technology to Level 3 automated driving – a more advanced classification in which the car can take over the driving task, but not all of the time, under specific conditions. Even though level 3 systems are in operation, there are still limitations to its use and the driver must be prepared to take control when needed.
As Hyundai advances to higher levels of automation, Park said it’s important to utilize multiple sensor types.
Park said that, in order to guarantee safety, sensor redundancy is essential; in the event that one sensor fails, other sensors must be able to operate safely.
There is no clear plan yet for commercialising Hyundai’s Level 3 technology. It is also indicative of the industry’s general woes as it tries to progress from driver-assistance systems to those that can tackle more complex driving scenarios.
The difference between the various stages of automation is especially significant. Level 2 systems may help to steer the car, accelerate and decelerate the car, but the driver will be responsible for maintaining road and vehicle awareness. Generally, Level 2+ indicates more advanced assistance, such as highway driving, and Level 2++ may be given extra assistance in more complex situations, such as urban driving. Although these systems have more capabilities, they still need driver participation.
The technology race isn’t just about the number of sensors you can fit or how much computing power your processors have. Artificial intelligence is crucial to modern driver assistance systems, which are fed vast amounts of practical driving data during training. This has led to data collection being one of the most valuable resources for automakers who want to enhance their software.
Hyundai and Kia think they can have a lot of impact in this field, having a presence on their own in both regions. The two firms sell over 7 million vehicles per year, and between them have access to a potential massive amount of driving data. Hyundai plans to leverage information it gathers from its cars to improve its autonomous-driving capabilities.
Hyundai says that by 2033, the volume of their vehicles should enable them to outpace rivals in the amount of driving data they collect. Once reached, that goal could be a crucial aspect of the company’s strategy, as a self-driving system gets smarter with every time it sees new road conditions, traffic and unusual circumstances.
The Nvidia tie-up is more than an assist to the driver. Hyundai Motor Group is also strengthening ties with the tech firm in other areas, such as artificial intelligence data centres, as well as humanoid robots created by Boston Dynamics, which is owned by Hyundai.
