HITTER: A HumanoId Table TEnnis Robot Via Hierarchical Planning and Learning
Zhi Su, Bike Zhang, Nima Abraham Rahmanian, Yuman Gao, Qiayuan Liao, Caitlin Regan, Koushil Sreenath, Shankar Sastry
AI summary
Problem
Humanoid robots struggle with rapid, precise interaction in dynamic environments due to extreme time constraints and the need for coordinated whole-body agility. Table tennis exemplifies this gap, requiring sub-second perception, prediction, and balance recovery during high-speed exchanges.
Approach
The system separates high-level trajectory prediction and strike planning from low-level control, using a model-based planner to forecast ball dynamics and an RL-trained whole-body controller that mimics human motion references to execute agile strikes and maintain stability.
Key results
- Model-based planner achieves <7.5 cm position and <20 ms timing error before impact
- RL controller reaches targets within 0.75 m in under 0.8 s with 94.3% success rate
- Real-world deployment sustains up to 106 consecutive shots against a human opponent
- Autonomous humanoid-humanoid rallies demonstrated without specialized hardware
Why it matters
Validates agile, interactive whole-body control on general-purpose humanoids, advancing robotics for high-speed dynamic tasks and human-robot collaboration.
Abstract
Humanoid robots have recently achieved impres- sive progress in locomotion and whole-body control, yet they remain constrained in tasks that demand rapid interaction with dynamic environments through manipulation. Table tennis exemplifies such a challenge: with ball speeds exceeding 5 m/s, players must perceive, predict, and act within sub-second reaction times, requiring both agility and precision. To address this, we present a hierarchical framework for humanoid table tennis that integrates a model-based planner for ball trajectory prediction and racket target planning with a reinforcement learning-based whole-body controller. The planner determines striking position, velocity and timing, while the controller generates coordinated arm and leg motions that mimic human strikes and maintain stability and agility across consecutive rallies. Moreover, to encourage natural movements, human motion references are incorporated during training. We validate our system on a general-purpose humanoid robot, achieving up to 106 consecutive shots with a human opponent and sustained exchanges against another humanoid. These results demonstrate real-world humanoid table tennis with sub-second reactive control, marking a step toward agile and interactive humanoid behaviors.