Robot arm serves a cereal breakfast

Nov 10, 2026 Last reply: 3 hours ago 3 Replies
Jump to replies (3)

The hardest part of this build is plainly getting an arm to handle breakfast foods without wrecking the table, and the maker tackles it head on. Cereal ends up scattered, the box gets knocked off the table edge, milk splashes over the bowl, and the spoon drops once feeding is done. Even with all that, the machine runs its whole routine from start to finish, mess and all. That honesty about the failures is what makes the project charming to me.



The hardware centers on a uArm robot arm with its own Arduino board, driven by servo motors with exposed wiring. A suction cup on the arm picks up objects, and the whole setup sits on a rectangular base secured with duct tape on a wooden table. The maker describes the uArm as pricy but very nice to work with, and the smoothness of the arm's movements supports that view in my eyes.



The programming is done through the arm's learning mode. A button press switches the suction cup on and off, and a recorded sequence can be replayed at any time with another press. The routine itself is tidy: the arm rotates and tips the cereal box over a bowl, then lifts and tilts the milk carton before letting go, and finally grasps a pink plastic spoon, lowers it into the bowl, and raises it to the maker's mouth while a paperback book is being read.



One thing stays unresolved. The maker is not sure whether learning mode can loop a sequence over and over, and the sequence is not stored, so it vanishes when the arm is powered down and back up. Developers are reportedly working on translating learning mode movements into Arduino code, which in my opinion would close that storage gap nicely.



Project by Simone Giertz. Watch it on their YouTube channel:

formatting link


Teaching the routine through learning mode instead of writing code seems like a smart shortcut to me. The button toggle for the suction cup keeps the teaching session simple, but the trade-off shows up in storage: once the arm loses power, the whole recording is gone. Coding it in Arduino would take longer up front yet survive a reboot.

One thing I keep wondering: since the recorded sequence is not stored, does the entire routine have to be re-taught after every power cycle, or does part of it survive somewhere? Redoing a full teaching session before each breakfast would get old fast, so I hope there is a workaround.

Here is a variation worth trying: once developers finish translating learning mode movements into Arduino code, the sequence could live on the board and survive power loss. Adding a loop would also let the arm repeat the feeding without anyone pressing the replay button each round, which would suit a slow reader at the table.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required