Why Camera-Only Autonomy Hits a Wall for Trucks

A camera-only self-driving car is like ChatGPT on wheels – dazzling when it works well, but not immune to hallucination. For the trucking industry, the camera-only concept is a fantasy that ends in a crash course in physics. I’ll explain why.

For any autonomous driving system, knowing how far away objects are and their relative velocity is critical. A car doing 70 mph needs roughly 100 metres to stop in an emergency, and out to that distance, camera-based depth estimates are generally good enough. (The keyword there is “estimates”, because they are not measurements, but educated guesses.)

Camera systems estimate depth by triangulation. That means – as anyone with two eyes can tell you – the further away something is, the harder it is to precisely gauge its distance or speed. It’s an unavoidable mathematical reality that with triangulation, uncertainty doesn’t grow linearly, but quadratically. In other words, double the distance and you quadruple the error.

At a range of 200 metres – the distance a fully-loaded Class 8 truck travelling at highway speeds needs to stop – you simply can’t trust cameras any more. For example, a camera-based system might be accurate to within 8 meters at a range of 100m, but off by more than 30m at a range of 200m. That’s the difference between stopping safely and ploughing through an obstacle whose position and speed you misjudged.

The irresistible solution is to add LiDAR. LiDAR doesn’t suffer from this dramatic gain in uncertainty. It doesn’t infer distance, as camera systems must, but measures it directly by sending out pulses of light and timing how long they take to bounce back from whatever surrounds the vehicle. And LiDAR’s uncertainty increases only linearly – from a few centimeters at 100m to just a few more centimeters at 200m. That’s the kind of precision you can build a safety case on.

In short, cameras interpret the world, while LiDAR measures it – the eyes and the tape measure working together. And some modern LiDAR sensors go further still, capturing not just distance but also relative velocity, much like radar can. The price of LiDAR units has fallen dramatically in the last decade, so why wouldn’t you want it in your sensor mix?

At PlusAI, we’ve tested both approaches. Turn off the LiDAR and everything still looks fine up close, but at long range the picture falls apart. Once you’ve seen that difference, you can’t unsee it.

There’s a second safety argument to consider. Camera-based self-driving systems rely on neural networks trained on millions of examples of driving scenarios. They learn by imitation, not by rule, which means you can never prove mathematically that they’ll behave safely in every situation. A Matchbox car held close to the lens might look like a full-sized sedan in the distance, because the model assumes cars are usually about 1.8 metres wide. The only way to build confidence in camera-only systems is to log billions of real-world miles and hope to capture as many unusual driving scenarios as possible.

That’s why a company like Tesla, with its enormous fleet constantly feeding data back to headquarters, can at least attempt a camera-only approach. Systems incorporating LiDAR simplify that statistical safety argument dramatically. They don’t infer distance – they measure it – with physical precision that can be tested, validated, and trusted.

This difference speaks to more than technology; it’s about philosophy. Camera-only self-driving vehicles fall into the realm of recreational autonomy – a bit like skiing, where the thrill comes with risk, and you might break your legs (or worse). Autonomous trucking isn’t recreational. It’s industrial-grade autonomy, more akin to a pacemaker: something that must work safely, predictably, and every time. When an 80,000-pound vehicle is moving at highway speed, “good enough” isn’t good enough.

A self-driving truck should live within its comfort zone, not at the edge of its capability. The safest system is one where perception is effortless and knowing what’s out there is simple, direct, and unambiguous. Because for autonomous trucks there’s no place for hallucination, only measurable truth. And that’s what LiDAR delivers.