DexDrummer: In-Hand, Contact-Rich, and Long-Horizon Dexterous Robot Drumming
2026-03-23 • Robotics
Robotics
AI summaryⓘ
The authors introduce drumming as a new way to test robotic hand dexterity since it requires precise finger control, frequent contact with drumsticks and surfaces, and long sequences of coordinated actions. They developed DexDrummer, a system that learns to play drums using a combination of planned movements and fine-tuned adjustments, trained first in simulation and then applied to real robots. Their method handles complex finger interactions and quickly switches between drums, outperforming simpler approaches. In both simulated and real settings, DexDrummer successfully played multiple drum styles with high accuracy.
in-hand manipulationdexterous manipulationreinforcement learningsim-to-real transferbimanual controltrajectory planningcontact-rich dynamicsrobotic drumminghierarchical policy
Authors
Hung-Chieh Fang, Amber Xie, Jennifer Grannen, Kenneth Llontop, Dorsa Sadigh
Abstract
Performing in-hand, contact-rich, and long-horizon dexterous manipulation remains an unsolved challenge in robotics. Prior hand dexterity works have considered each of these three challenges in isolation, yet do not combine these skills into a single, complex task. To further test the capabilities of dexterity, we propose drumming as a testbed for dexterous manipulation. Drumming naturally integrates all three challenges: it involves in-hand control for stabilizing and adjusting the drumstick with the fingers, contact-rich interaction through repeated striking of the drum surface, and long-horizon coordination when switching between drums and sustaining rhythmic play. We present DexDrummer, a hierarchical object-centric bimanual drumming policy trained in simulation with sim-to-real transfer. The framework reduces the exploration difficulty of pure reinforcement learning by combining trajectory planning with residual RL corrections for fast transitions between drums. A dexterous manipulation policy handles contact-rich dynamics, guided by rewards that explicitly model both finger-stick and stick-drum interactions. In simulation, we show our policy can play two styles of music: multi-drum, bimanual songs and challenging, technical exercises that require increased dexterity. Across simulated bimanual tasks, our dexterous, reactive policy outperforms a fixed grasp policy by 1.87x across easy songs and 1.22x across hard songs F1 scores. In real-world tasks, we show song performance across a multi-drum setup. DexDrummer is able to play our training song and its extended version with an F1 score of 1.0.