A humanoid robot now plays drums with human-like skill, mastering songs and rhythms through AI training. Read on!
Researchers at SUPSI, IDSIA, and Politecnico di Milano have introduced Robot Drummer, a humanoid robot capable of playing drums with accuracy and expression using a reinforcement learning algorithm. In simulation tests on the G1 humanoid robot from Unitree, it achieved rhythmic precision exceeding 90%, performing complete musical tracks while developing human-like drumming behaviors such as characteristic movements, stick switching, arm crossing, and rhythm adaptation.
The system played popular songs from jazz, rock, and metal genres — including In the End, Take Five, and Living on a Prayer — and learned advanced strategies like planning future strikes and adjusting stick use mid-performance. These abilities emerged entirely from the reward-based training process.
At its core, the system represents each piece of music as a sequence of precisely timed contact events called rhythmic contact chains. These specify which drums to strike and when. Practicing in a simulated environment allowed the robot to refine its drumming techniques and optimize movements to meet different rhythmic demands, similar to how human drummers adapt their playing.
The idea for Robot Drummer came from an informal conversation about how humanoid robots, while increasingly capable in a wide range of tasks, rarely take part in creative or expressive domains. Drumming was selected as the challenge because it is rhythmic, physical, and requires rapid coordination across multiple limbs.
The study’s goal was to develop a machine learning–based system that would enable a humanoid robot to perform complete songs with human-like timing. The team now plans to transfer these learned drumming skills from simulation to physical hardware for live performances. Future development aims to give the robot the ability to improvise and respond to musical cues, moving beyond fixed drum patterns toward a more human-like capacity to “feel” and adapt to the music.
