Organisation
Multiple US research institutions
Year
2026
Substrate
Brain organoid
Architecture
Reservoir computing + robotic control
Task
Object grasping and laser chasing
Evidence
Preprint

Brainobot uses a brain organoid as a high-level decision layer inside a closed robotic control loop. Sensory information is encoded for the organoid, neural activity is recorded, and the decoded output contributes to motor actions on a humanoid robot.

The September 2026 bioRxiv preprint reports proof-of-concept demonstrations including object grasping and laser chasing. The authors describe the organoid layer as reservoir computing and report cross-task adaptation, high computing efficiency and low energy consumption.

Evidence status

Preprint research demonstration. Performance and energy claims still need peer review, matched baselines and independent replication.

A physical sensor–decision–action loop

Previous organoid-computing experiments have mainly used virtual tasks, classification or reservoir readouts. Brainobot places an organoid inside a physical sensor–decision–action loop, where changing inputs can affect a robot operating in the world.

No matched comparison yet shows an organoid controller outperforming conventional robotics hardware. The experiment demonstrates an architecture that inserts living neural tissue into a physical embodied-control stack.