The hard part of a de-palletizing robot isn't the picking, it's the cage

This cage unloading and stacking work sounds lighter than loading and unloading goods, but it may not actually be. Cage carts have never been standardized in the industry.

I've recently been looking at Sainade's line. The iLoabot series now has three models: M for autonomous loading and unloading, X for autonomous operations, and P specializing in cage unloading and stacking. At the Industry Fair, P demonstrated a tracked chassis overcoming obstacles in scenarios with height differences. Its default working environment is the realistic site with uneven ground and randomly placed cages, not a flat laboratory floor.

Being willing to use a tracked chassis shows they understand that cage carts run all over the place, not stay at a fixed workstation waiting for someone to serve them.

But what's more worth looking at is what's behind it. Sainade equipped this system with a multimodal VLM large model, following the chain of perception, decision-making, and execution. In the official wording, what's repeatedly mentioned is many SKUs, messy stacking patterns, inconsistent standards. These three phrases are the real pain points. The stacking action itself isn't hard; what's hard is that what's stuffed into the cage is different every time—last cage was cardboard boxes, the next cage might be a pile of soft packages.

In March this year, Sainade received over 200 million yuan in strategic financing, with Sinotrans, Alibaba, and Fortune Capital all on the list. Capital being willing to bet shows that the last 20 meters of loading and unloading is indeed being treated as a tough nut. But financing doesn't equal implementation. My advice is still the same old thing: don't just look at the demo on the exhibition stand; go see what it looks like after running at a customer site for three months, especially long-tail situations like cage cart changeovers and cargo misalignment.

Last month I wrote an on-site observation of force-controlled dual arms; the logic is the same—the parameter sheet looks good, but only real working conditions reveal the truth.

For non-standard carriers like cage carts, is it about making robots adapt to them, or should the industry first standardize them before talking about scale? There's no answer yet.