L4 Warm-Up Before the Wuzhen Summit: MINIEYE Gained More Than Just a Shuttle Order

Before the Wuzhen Summit opened on November 2, 2026, I was going through bid award notices and came across a procurement deal nobody was talking about: Tongxiang quietly deployed six L4 autonomous minibuses to provide free shuttle service during the summit, running all the way from the Expo Center to Puyuan Scenic Area.

The winner wasn't Baidu or Pony.ai — it was a company called MINIEYE, Hong Kong stock code 2431, with a contract worth 11.85 million yuan.

Most people haven't heard the name. But this story is far more interesting than those big slide decks at product launches.

This Is Nothing Like Apollo Go

When you hear "L4," you probably picture something like Apollo Go — driverless taxis roaming the streets.

These six vehicles in Tongxiang are a completely different game.

Fixed routes, just a handful of stops, low speed, and a safety operator still on board. It's basically like the sightseeing shuttle you'd ride in a park — except the steering wheel turns itself.

MINIEYE iRobo series autonomous minibus product image

So why target such a seemingly "low-tech" scenario?

Think about it and it makes sense. How much money is Robotaxi burning right now? Tens of thousands of yuan in sensors per vehicle, fleets of hundreds, costs that won't come down, and commercialization still nowhere in sight.

But small scenarios like scenic area shuttles and campus circulators are different. Routes are fixed, foot traffic is controllable, speeds are just 20–30 km/h, and the technical difficulty drops by an order of magnitude. Most importantly — governments and scenic areas are actually willing to pay.

11.85 million yuan isn't big money, to be honest. But it's a full package: six vehicles, three routes, roadside perception, and ongoing operations and maintenance. What's really worth thinking about is whether other places will copy this playbook once Tongxiang's system runs smoothly.

How Did an ADAS Tier 1 End Up Building Autonomous Minibuses?

Now let's talk about MINIEYE. It's actually quite well known in the automaker supply chain.

Founded in Shenzhen in 2014 by Liu Guoqing, a 1987-born PhD from Nanyang Technological University. Its main business is selling ADAS solutions to automakers — from automatic emergency braking and lane keeping to highway navigation and urban NOA. Its customers include NIO, Neta, FAW, and Geely, and it has already reached mass-production partnerships with over 40 automakers.

Roof-mounted surround sensor array: 3 primary radars + 3 blind-spot radars

Put simply, it's a Tier 1 in the automotive supply chain, walking the same path as Horizon Robotics. The difference is Horizon sells chips, while MINIEYE sells algorithms plus complete systems.

So how does a company that supplies automakers end up making minibuses?

Once you think it through, it's not surprising at all. After years of doing L2+ for automakers, what is it best at? Automotive-grade reliability, cost control, and mass-production delivery. Transplant all of that directly onto L4 minibuses, and it fits perfectly.

Companies like Baidu's Robotaxi division are strong on technology, but there's still a massive gap between what they build and something that can be mass-produced, made affordable, and run reliably.

In 2026, the six vehicles in Tongxiang received upgrades — top speed increased by over 30% compared to 2025, and those jarring sudden brakes are basically gone. It now drives like a seasoned veteran. Three primary radars plus three blind-spot radars wrap around the vehicle body, centimeter-level positioning, and it handles crowded intersections steadily.

Three Signals More Substantive Than Any Launch Event

It's not a big deal, but after watching it for a while, there are a few things I want to highlight.

What surprised me most is that L4 deployment is happening faster than you'd think. When people hear "autonomous driving," they assume it's still ten or twenty years away — but it's already running in scenic areas, ports, and campuses. Not lab demos, but real operations carrying passengers and generating revenue.

Then there's the question of who's making money. When people think of autonomous driving, all eyes are on Baidu, Pony.ai, and WeRide. But quietly profiting are the Tier 1 ADAS suppliers. Automaker orders are their bread and butter, and L4 scenario deployment is new incremental growth — they win on both fronts.

I also took a closer look at the Tongxiang template — whether it can be replicated is the key question. The Yangtze River Delta has plenty of scenic areas, well-funded local governments, and high receptiveness to new technology. If this shuttle data looks good, Jiaxing, Suzhou, and Huzhou will likely follow suit.

Of course, I'm not saying this is a huge deal. 11.85 million yuan, for a company with billions in annual revenue, is far more symbolic than material. L4 is still at a very early stage — one minibus on the road doesn't mean the whole industry has made it.

But the direction is right. Look at those launch events — each PPT more grandiose than the last, each vision more distant. It's these minibuses quietly running on real streets that look more like what technology deployment should actually look like.

At the end of the day, L4 isn't about whose algorithm computes faster — it's about who can first make the numbers work in a specific scenario.