Robot mowers don't need parking spots, but they still fear breakpoints

The PANDAG G1's capabilities are real, but it's more like a tool for reducing headcount on landscaping crews, not something you buy and bring home to save yourself trouble. My judgment is it depends: for large, regular green areas it's worth it; for fragmented terrain, places with lots of trees, and poor signal, don't expect it to cover for you.

I came across it by chance. At CES 2026, Shitieshou had the G1 on display, and I'd originally gone to look at cockpits, but got stopped by that thing over a meter wide. Later a friend doing park landscaping also sent me a video he shot at Equip Expo, plus the on-site deployment footage the salesperson gave me, so I got a pretty good look at the whole process. Not the kind of thing I brought home to test—let me be clear about that.

The official line is RTK plus 4G plus lidar plus vision, four systems stacked together. Anyone who works on parking would smile at this combo—it's the classic "one set isn't enough, throw in redundancy." RTK gives absolute coordinates, centimeter-level, provided the sky can see satellites; lidar handles nearby obstacles and boundaries; vision adds semantics, able to tell grass from a flowerbed edge.

The onboarding process is similar in logic to a robot vacuum, but the bar is much higher. Step one is mapping: a person walks around the site with a remote or phone dropping points, marking the mowable area, no-go zones, and dock location. I watched an engineer operate it at the booth—the phone interface is a top-down view with divided zones, colors clearly distinguished, and drawing a zone takes a few seconds. Step two is setting the route; the G1 plans a reciprocating path itself, with a 48-inch deck, one pass covers over a meter, and for large areas the efficiency really does beat manual labor.

Step three is where the real skill shows: unattended loop operation. It finishes mowing, returns to the dock itself, and comes out again the next day per schedule. This is meaningful for golf courses, solar farms, and large parks, because the common traits of these sites are large area, regular terrain, and clear boundaries.

The pitfalls are here too.

One is tree shade. RTK is steady as a nail on completely open lawn, but once under a canopy or walking close to buildings, the fixed solution easily drops to a float solution, and after losing the fix it slows down, hesitates, even stops and waits for a person. This is the same class of problem as no signal in an underground garage—except one is no signal, the other is signal blocked. The salesperson told me it now relies on lidar and vision to keep going; I believe half of it—being able to keep going and going accurately are two different things.

The second is that "can mow 6-foot-tall grass." 6 feet is nearly 1.8 meters, and I'm skeptical of that number. My understanding is it refers to the extreme condition of a first cut on wasteland, not a routine maintenance height. At that height the grass is lodged, and the deck getting through doesn't mean getting through cleanly—most likely you still have to go over it again. Between marketing and reality, this kind of number is the easiest to inflate.

The third is modularity. Edge-trimming, spraying, and other expansion modules are selling points, but after swapping them you have to recalibrate, and my friend's feedback is that one calibration takes half a day. Modularity saves the money of buying a second machine, but costs people's time.

It suits golf courses, large parks, solar or orchard sites that have their own operations teams—large area, clear boundaries, and able to accept the upfront investment in mapping and calibration. Private gardens, fragmented terrain, lots of trees, corners with poor signal, and properties that want to completely wash their hands of it—don't count on it. It's a tool that needs people to manage it; don't expect to fully let go.

Commercial mowing robots will probably shift from comparing "how clean it mows" to comparing "whether it can pick up on its own after an interruption." This is the same road parking went down: the first generation compared who could park in, the second compared who could keep parking after being blocked. The mowing industry is still in the earlier stage.