Hardware·
MIT Unveils Featherweight Flying Robot With Onboard Sensors and Compute
A new flapping‑wing microrobot from MIT carries a 244 mg electronics package, achieving a lift‑to‑weight ratio of 1.9 and demonstrating autonomous navigation without external motion capture.

MIT researchers have unveiled a flapping‑wing microrobot that integrates its own sensors and a microcontroller, removing the need for an external motion‑capture system.
The robot achieves a lift‑to‑weight ratio of 1.9, giving it enough thrust to carry a 244‑milligram electronics package that handles onboard sensing, state estimation, and control.
In outdoor tests the vehicle followed a complex trajectory and landed precisely on a sunflower, proving that autonomous flight is possible at the sub‑gram scale.
Why it matters for GPU / AI infrastructure
The demonstration highlights ultra‑low‑power edge AI workloads — tiny neural‑network inference running on a microcontroller — a class of compute that benefits from the high‑density GPU clusters AiGpu provides for model training and large‑scale inference.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- microrobot
- edge ai
- precision agriculture
- hardware
By AiGpu Editorial · Editorial rewrite based on public reporting (MIT News AI)
← All articles