The global push toward fully autonomous driving has historically faced a major geographic and economic barrier. For years, Level 4 (L4) vehicle development-where the car handles all driving functions within a specific area without human intervention-was concentrated in highly structured urban environments across North America and China.
These solutions were based on the deployment of costly sensors and compute stacks, which made the vehicles prohibitively expensive to become mainstream solutions. Besides, conventional autonomous vehicle software does not perform well due to the exposure to diverse driving conditions in emerging economies.
VinFast, a Vietnamese electric vehicle manufacturer, along with Autobrains, an AI mobility company, have now come up with a solution to this problem, according to an announcement made during the NVIDIA GTC Taipei event at COMPUTEX.
Under this collaboration, the two companies seek to create a next-gen Level 4 robotaxi and production vehicle program for use in Southeast Asia. With the pairing of Agentic AI software developed by Autobrains with NVIDIA DRIVE Hyperion™ 10 compute, the two companies are creating an affordable, scalable solution for autonomous vehicles.
A Modular Approach to Complex Road Conditions
The collaboration introduces a tailored approach to autonomous software architecture, moving away from traditional “end-to-end” deep learning models that require continuous, massive processing power. Instead, the program focuses on modular, situational processing.
Key structural components of the Level 4 program include:
Agentic AI software by Autobrains: In contrast to conventional systems that operate large models at all times, Agentic AI makes use of sub-agents that operate independently of each other. The sub-agents are activated only on occasions where there is a need to do so, for example, when dealing with a group of two-wheelers or an unidentified intersection.
NVIDIA DRIVE Hyperion 10 Platform: The hardware foundation utilizes NVIDIA’s pre-validated DRIVE Hyperion 10 compute architecture. This standardized tier provides the necessary processing muscle for real-time sensor fusion while reducing the long validation timelines that typically delay AV programs.
Compatibility with Standard Sensors: The system has been designed with a view to meeting the level 4 standards by utilizing standard automotive camera and radar sensors without necessitating the use of expensive sensor systems.
Also Read: MediaTek and NVIDIA Announced Partnership Reshape Premium Silicon with RTX Spark
Validation Hub in Local Context: Validation of the system through real-world testing is currently taking place via the use of VinFast’s high-end VF 8 and VF 9 all-electric SUVs within the city of Hanoi.
Impact on the Automotive Industry
The three-way partnership between VinFast, Autobrains, and NVIDIA signals a major shift in how the Automotive sector approaches vehicle intelligence:
1. Shifting Autonomy from Luxury to Mass Market
Previously, high-end driving autonomy has been associated with luxury car prices. VinFast’s vision goes contrary to this status quo by using the modular design to reduce the processing needs of autonomous vehicles, thus creating accessible self-driving capabilities. This move puts pressure on other manufacturers based in the West and China to reconsider their premium business models concerning ADAS.
2. The Emergence of a Self-Driving System for the Region
Most autonomous vehicles have been tested within highly-organized and predictable driving environments. Through its partnership, VinFast proves that autonomous vehicles are capable of navigating through dynamic and busy driving situations in the ASEAN region. The AI software developed to operate such vehicles is highly flexible and robust, which makes it viable for exports globally.
3. Reimagining the Tier 1 Software Supply Chain
Through this partnership, we see an increasing trend where the auto manufacturer works closely with specialized AI software providers as opposed to relying only on Tier 1 suppliers. Licensing the software from Autobrains while combining it with computing hardware provided by NVIDIA, the automaker can keep the integration and regional tuning capabilities internally.
Overall Effects on Businesses Operating in the Sector
The following points explain the strategic benefits of the Level 4 concept for stakeholders in the automotive industry:
Autonomous Robotaxis Financial Sustainability: Prior to this, the high hardware prices had made financial sustainability difficult for the automated rideshare business model. However, using affordable Level 4 vehicles, it becomes easier for regional transportation service companies to set up their profitable robotaxi fleets, taking the first step towards establishing a fully autonomous mobility ecosystem.
Scaling Up the Value Chain for New EV Companies: In case of new EV companies, competing through a price war on the basis of battery features or quality makes no sense. However, due to the shift to software-led autonomous systems, VinFast and other brands can ensure their own survival through the creation of differentiating ecosystems.
Boosting Demand for High-End Hardware: As vehicles become software platforms, the demand for advanced hardware is likely to increase. In this case, semiconductor and automotive optics manufacturers are likely to stand to gain from it.
Conclusion
“Higher-level mobility solutions are not supposed to become luxurious,” Prof. Dr. Nguyen Van Duong, the Deputy CEO of ADAS at VinFast Global. The joint venture between Autobrains and NVIDIA is a clear reflection of this sentiment, which brings about an efficient and budget-friendly solution to the challenges faced in modern automobile automation. Through combining adaptive software and reliable computer hardware, the companies are proving that Level 4 autonomous driving is attainable without breaking the bank. As far as the auto industry is concerned, this initiative proves that the success of autonomous driving lies in execution.
