On a São Paulo trade-show floor this week, a 50-centimeter-tall humanoid named M1 is supposed to do something that sounds mundane and turns out to be the whole product. When its battery gets low, it rolls itself across the booth, lines up against a small wheeled cart, and recharges, with no human fetching the cable and no human telling it where to go. The company showing it, Chinese robotics startup Zeroth, is calling the M1 the world's first home humanoid with a fully autonomous energy closed-loop. The loop, in the company's framing, runs from "I need power" to "I am powered" without help.
Whether that loop actually works in somebody's living room, as opposed to a choreographed booth demo, is the only question the rest of the announcement has to answer. The Eletrolar Show appearance, the second product, the regional rollout pitch: all of it is interesting mostly as context for whether the self-recharge claim survives contact with a real home. Everything below either sharpens that question or distracts from it.
Zeroth, the company's romanized name (written 元点Zeroth in Chinese), is a Chinese consumer-robotics startup in the broader category that Chinese industry calls 具身智能, often translated as "embodied AI." That phrase, which the rest of this piece uses sparingly, means robots meant to act in the physical world rather than only on screens. Other humanoid robotics efforts, from Tesla's Optimus to Figure's 01, 1X's Neo, Unitree's H1, and Sanctuary's Phoenix, exist in the same broad category, but Zeroth is pitching a much smaller device: a roughly knee-high companion meant for the living room, not a factory floor. At Eletrolar Show 2026, Latin America's largest consumer-electronics trade show, running this week at São Paulo's Distrito Anhembi convention complex, the company is making its first Latin American appearance.
The "world's first" framing is the company's, not an independent benchmark. Founder and CEO Guo Renjie (郭人杰) framed Latin America as a "pre-transformation market" where the company said embodied AI can reshape daily life; the wider claim, that the M1 is the first home humanoid with a fully autonomous energy closed-loop, was attributed to the CEO in a Chinese tech outlet's coverage of the company's announcement. That distinction matters because the strongest version of the claim, that no other home robot on the market today can dock itself without human help, is a falsifiable product statement, not an award. It is also a claim that competitors in the same broad category (humanoid or otherwise) have, in some form, attempted: robot vacuums have returned to a dock for years, and the harder question is whether a 50-centimeter biped can do the equivalent with a small auxiliary cart.
So what does the loop actually do, and what would it take for the claim to mean what Zeroth says it means? Three steps have to happen without human input. The robot has to notice its own battery is low, decide where the charger is, and physically complete the docking: navigate across a room, line up against a small wheeled cart that holds the charging contacts, and engage. Each step is a different engineering problem. Battery monitoring is the easiest, since every robot has done this for years. Localization and path planning in an unmapped home is harder, and it is the same problem robot vacuums have been working on for a decade, only now with a small bipedal robot instead of a puck. The mechanical docking step is the genuinely novel piece, because a humanoid is not shaped like a Roomba: it has to find a small target, position itself against it, and engage without tipping over.
What the company has shown so far, based on the Chinese tech media coverage of the launch event and Brazilian tech outlet TechTudo's reporting from the Eletrolar floor, is a working prototype on the booth floor. What it has not shown publicly is the loop surviving a home environment, with furniture, pets, cables, and a charger that has been moved a meter to the left since the last dock. That is the gap between "world's first at a trade show" and "world's first in a living room." No price, no shipping date, and no independent home-deployment footage appear in the available English-language or Chinese coverage of the Eletrolar announcement; that absence is itself a data point.
The second product on display, the W1, is a tracked mobile robot, basically a low rectangular base on tank-style treads, with a screen or sensor head on top. The company's US launch release, the cleanest English-language write-up of the lineup, positions the W1 for both home and light-commercial use: grass, gravel, mild slopes, indoor-to-outdoor transitions, voice interaction, autonomous navigation, and the ability to be dispatched remotely. That is a useful set of features for a multi-surface patrol robot. It is also, as a category, much closer to incumbent products such as wheeled delivery robots, robotic lawn equipment, and security patrol bots, and harder to claim as a "first."
The bigger context is that this is not a wholly new product. At CES 2026 in January, the company laid out a roadmap called Jupiter that ran from a compact M1 companion toward a full-size teleoperated and eventually autonomous humanoid, which The Verge also covered at CES 2026. Brazil is the next step in a US-then-Latin-America sequencing, and the M1 and W1 lineup is the same one that debuted in the US in late 2025. So the question of whether the M1 can actually dock itself in a home is also a question of whether the company's roadmap is closer to shipping than to demoing, and the Eletrolar appearance, while new for the region, is mostly a stage for showing that the loop works, not for proving it does.
The honest version of the Eletrolar story is also the most useful one for anyone deciding whether to pay attention to the next "world's first home robot" announcement. The mechanism, not the region, is the test. If the M1 can dock itself in a real apartment, with a charger that someone has moved, and do it without a human watching, then the energy closed-loop is a real autonomy milestone, and the rest of the category has a harder benchmark to clear. If it can't, then "world's first at a trade show" is exactly what it sounds like: a press-release line waiting to be falsified. The company is saying the answer is the first one. Until the second one is independently demonstrated, that is a claim, not a result.