UCLA researchers have unveiled a $400 two-legged robot that carries a smartphone, follows a person, and positions itself for photos, a low-cost platform they pitch for classrooms and rapid prototyping.
Scientists at the University of California, Los Angeles have demonstrated PhoneBot, a compact two-legged machine built around an ordinary smartphone. The device trots alongside a person, tracks its target, stands up from a face-down position, and can frame a shot with the phone’s camera.
Daily Mail reporting describes the platform as roughly 19 inches tall and about four pounds, with a stated cost of $400. The phone sits above the “hips” and forms the torso.
Initial demonstrations showed basic walking, person tracking, and recovery from a prone position. The phone’s camera spots and follows people. Its motion sensors feed orientation data. Software on the phone sends commands over Wi-Fi to a small onboard computer that drives the motors.
PhoneBot uses 13 motors, six in each leg and one that turns the torso. That setup lets the camera swing sideways while the feet stay planted. The team tested a range of phones. All of them handled core sensing, movement control, and vision tasks.
In their preprint posted to arXiv, the researchers stated the design goal plainly.
"This paper presents PhoneBot, a low-cost, open-source compact humanoid platform that integrates a commodity smartphone with a modular body for sensing, on-board computation and interaction."
They describe smartphones as “a practical basis” for robotics and say they want “an affordable, reproducible platform for education, research and rapid prototyping.” The open-source approach aims to put a working humanoid in the hands of students and labs that cannot spend tens of thousands on commercial robots.
Right now the machine has no arms. The team plans to add lightweight limbs so it can handle objects. Testing has stayed on flat terrain. Uneven surfaces remain on the to-do list.
Spoken commands already work through the phone’s microphone and speaker. Speech recognition and language processing still run on a separate laptop. The researchers hope to move those AI models onto the phone itself.
One suggested use is a roaming photographer that keeps everyone in the frame without a human holding the camera. The same low price and modular body could let classrooms and small labs experiment with bipedal robots without a big budget.
PhoneBot is still an early research platform, not a consumer product. It does not climb stairs, carry packages, or run advanced AI onboard yet. What it does show is a clear engineering bet: strip the expensive custom sensors and computers out of a small humanoid and let a commodity phone do the sensing and high-level control.
That approach matches a broader pattern in American labs that favor open designs, off-the-shelf parts, and reproducible builds over closed, high-cost systems. A four-pound machine that walks, tracks, and rights itself for a few hundred dollars is the kind of incremental, testable progress that actually moves the field.
When researchers focus on cost, open hardware, and classroom use instead of flashy demos that never leave the lab, taxpayers and students get tools they can actually touch.