The Camera-Only Revolution: Can VinFast Crack Self-Driving Without Breaking the Bank?

VinFast just made a bet that could reshape the autonomous driving industry. The Vietnamese EV maker announced Tuesday it's partnering with AI firm Autobrains to develop self-driving technology that ditches the industry's most expensive component: lidar sensors.

While competitors burn through billions developing systems that cost $10,000-$50,000 per vehicle, VinFast is betting on a camera-only approach that could bring autonomous features to mass-market cars at a fraction of the cost.

The Lidar Dilemma

Traditional self-driving systems rely on lidar—laser sensors that create detailed 3D maps of surroundings. They're incredibly accurate but prohibitively expensive. Tesla famously abandoned lidar years ago, calling it a "crutch," while most others, including Waymo and Cruise, consider it essential for safety.

Autobrains takes a different approach entirely. Their system mimics human vision, using multiple cameras and AI to understand road scenarios without expensive hardware. The Israeli company claims their technology can achieve Level 4 autonomy—full self-driving in specific conditions—using only cameras and existing car sensors.

For VinFast, this partnership represents more than cost savings. The company, which launched just five years ago, is racing to establish itself in the global EV market dominated by Tesla and Chinese manufacturers. Affordable autonomous features could be their differentiator.

The Economics of Accessibility

Here's where the numbers get interesting. Current autonomous driving systems add $20,000-$100,000 to vehicle costs, limiting them to luxury models or specialized robotaxis. A camera-based system could reduce that to $2,000-$5,000, making self-driving features accessible in mid-range vehicles.

This cost reduction matters especially in emerging markets where VinFast operates. In Vietnam, Indonesia, and other Southeast Asian countries, even basic EVs remain expensive for most consumers. Autonomous features at current prices would be completely out of reach.

Autobrains CEO Igal Raichelgauz argues that human drivers navigate using only vision, so cars should be able to do the same. "We don't need to see in the dark better than humans or measure distances to the millimeter," he told industry analysts. "We need to understand situations like humans do."

The Safety Question Mark

But can cameras alone handle the split-second decisions that prevent accidents? Critics point to edge cases—scenarios where even slight sensor failures could prove catastrophic. What happens when cameras are obscured by snow, rain, or dirt? How does the system handle construction zones or emergency vehicles?

Tesla's camera-only approach has shown promise but also limitations. The company's Full Self-Driving beta has impressive capabilities yet still requires constant driver supervision. Recent data shows Tesla vehicles in Autopilot mode have accident rates similar to human drivers—better than some traditional systems, but not the dramatic improvement many expected.

Industry veterans remain skeptical. "Vision-only systems work until they don't," warns one autonomous vehicle engineer who requested anonymity. "The question isn't whether they work 99% of the time, but what happens in that critical 1%."

Market Disruption Potential

If successful, the VinFast-Autobrains partnership could force a industry-wide recalculation. Traditional automakers have invested billions in lidar-based systems. Ford, GM, and others might find themselves with expensive technology that suddenly seems obsolete.

Chinese EV manufacturers, already competing on cost, would likely rush to develop similar camera-only systems. This could accelerate the timeline for mass-market autonomous features from the current projection of 2030-2035 to potentially 2027-2028.

The partnership also highlights Southeast Asia's growing importance in automotive innovation. While most attention focuses on US and Chinese developments, companies like VinFast are quietly building capabilities that could leapfrog established players.