Microduck

The strongest positive reaction was not "cute robot" but "someone finally made the hobbyist robotics stack feel installable." People who had fought with Nvidia Isaac said the important part was getting a simulator and training flow working on a normal laptop in under an hour. That made the project feel less like a toy and more like a practical on-ramp for home robotics, especially because it appears to lean on the MuJoCo ecosystem that many already use for reinforcement learning and sim-to-real work. That, more than the duck form factor, is why technically minded readers paid attention. Skepticism clustered around the usual hardware traps. The battery life is only about an hour. Shipping, duties, and taxes push the real price above the headline number for many buyers. Several people treated the $395 price as surprisingly low for Dynamixel-based hardware, and one author reply bluntly called it a loss leader. Others immediately filed it under the long history of cute household robots that entertain for a week and then become shelf decor. There was also the standard concern that a camera-equipped robot aimed at homes and kids only makes sense if it can run fully offline. Replies pointed to Reachy and said the stack can run locally, including inference, which calmed some of that concern but did not fully settle it because the Microduck page itself was unclear. A lot of the noise in the comments came from side debates about robot pets, toy comparisons, and an AZERTY keyboard hint on the simulator page. The useful conclusion was simpler. People believe there is real demand for a low-cost, physically robust, locally programmable robot with a good simulator and clean SDK. They do not yet believe consumer robot pets are a durable category, and they do not trust preorder hardware timelines. So the current read is bullish on the tooling direction and cautious on the product category.