A team of researchers from Switzerland and Italy has developed a humanoid drummer capable of learning and playing complex rhythms live using reinforcement learning instead of pre-programmed sequences. Unlike traditional robotic systems, this drummer interprets drum sheet music as “rhythmic chains of contacts,” enabling natural human-like adaptations such as alternating hand usage or crossing hands while playing.
Training was conducted in a highly precise simulation using MIDI recordings from various musical genres, allowing the robot to build extremely accurate timing and expressive playing skills. The system coordinates up to 15 degrees of freedom in the upper body, planning movements across the entire drum set to maintain rhythm and spatial accuracy.
The result is impressive — the robot achieves over 90% rhythmic accuracy, matches skilled human drummers, and effortlessly performs demanding transitions and fast note sequences.
The next step is transferring these skills from simulation (tested on the Unitree G1 platform) to a physical robot capable of playing acoustic drums, performing live with bands, and improvising in real-time. This project is part of a broader trend moving from rigid programming to autonomous, adaptive learning in robotics.
In Brief: Tech World Highlights
- Adam Jonas from Morgan Stanley estimates that a humanoid rented at $5 per hour, replacing two workers paid $25 per hour, could be worth $200,000 in net present value.
- The new Fourier GR-3 is a humanoid “care-bot” standing 1.63 m tall and weighing 55 kg, equipped with visual, audio, and tactile sensors — including 31 pressure sensors, expressive movements, and a soft protective cover.
- Tesla is shutting down its Dojo supercomputer team, originally intended to advance autonomous driving and robotics, with team lead Peter Bannon leaving the company.
- New research in HCI/HRI shows that even a simple tabletop robot, initially designed for reading aloud, can become a significant emotional companion for children.
- A mobile robotic arm was tested on-site in Germany as part of human-machine collaboration, demonstrating teamwork between the robot and construction workers laying bricks.
AI Trending Tools:
- The Rundown – Product marketing.
- Grammarly – Compensation systems analyst.
- Databricks – Strategy and execution manager.