mlsanityData to 14 Oct 2026

PANDORA: Diffusion Policy Learning for Dexterous Robotic Piano Playing

arXiv 2503.14545 ↗First posted 17 Mar 2025cs.LG cs.RO cs.SDcited by 2

On the map: Diffusion models → Diffusion and flow policies for robotics

Abstract (arXiv)

We present PANDORA, a novel diffusion-based policy learning framework designed specifically for dexterous robotic piano performance. Our approach employs a conditional U-Net architecture enhanced with FiLM-based global conditioning, which iteratively denoises noisy action sequences into smooth, high-dimensional trajectories. To achieve precise key execution coupled with expressive musical performance, we design a composite reward function that integrates task-specific accuracy, audio fidelity, and high-level semantic feedback from a large language model (LLM) oracle. The LLM oracle assesses musical expressiveness and stylistic nuances, enabling dynamic, hand-specific reward adjustments. Further augmented by a residual inverse-kinematics refinement policy, PANDORA achieves state-of-the-art performance in the ROBOPIANIST environment, significantly outperforming baselines in both precision and expressiveness. Ablation studies validate the critical contributions of diffusion-based denoising and LLM-driven semantic feedback in enhancing robotic musicianship. Videos available at: https://taco-group.github.io/PANDORA

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models. Builds on 36 of its 40 references that are in the map. First posted on arXiv 17 Mar 2025.

Builds on

36 of 40 references.

Closest ancestors · 12

Same direction, newest first

  1. 2024DARE: Diffusion Policy for Autonomous Robot Exploration22 Oct
  2. 2024Diff-DAgger: Uncertainty Estimation with Diffusion Policy for Robotic Manipulation18 Oct
  3. 2024The Ingredients for Robotic Diffusion Transformers14 Oct
  4. 2024Diffusion Policy Policy Optimization1 Sep
  5. 2024Multimodal Diffusion Transformer: Learning Versatile Behavior from Multimodal Goals8 Jul
  6. 2024LDP: A Local Diffusion Planner for Efficient Robot Navigation and Collision Avoidance2 Jul
  7. 20243D Diffuser Actor: Policy Diffusion with 3D Scene Representations16 Feb
  8. 2023Diffusion Reward: Learning Rewards via Conditional Video Diffusion21 Dec
  9. 2023Zero-Shot Robotic Manipulation with Pretrained Image-Editing Diffusion Models16 Oct
  10. 2023EDMP: Ensemble-of-costs-guided Diffusion for Motion Planning20 Sep
  11. 2023Training Diffusion Models with Reinforcement Learning22 May
  12. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusion7 Mar

Other references · 24

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 681
  3. 2022Compositional Visual Generation with Composable Diffusion Modelscited by 624
  4. 2019MediaPipe: A Framework for Building Perception Pipelinescited by 583
  5. 2023Eureka: Human-Level Reward Design via Coding Large Language Modelscited by 571
  6. 2023Vision-Language Models are Zero-Shot Reward Models for Reinforcement Learningcited by 165
  7. 2024RL-VLM-F: Reinforcement Learning from Vision Language Foundation Model Feedbackcited by 156
  8. 2018Reinforcement Learning with Perturbed Rewardscited by 79
  9. 2024Exponential Moving Average of Weights in Deep Learning: Dynamics and Benefitscited by 71
  10. 2019Deep Reinforcement Learning for Industrial Insertion Tasks with Visual Inputs and Natural Rewardscited by 64
  11. 2023RoboPianist: Dexterous Piano Playing with Deep Reinforcement Learningcited by 30
  12. 2024DNAct: Diffusion Guided Multi-Task 3D Policy Learningcited by 23
12 more
  1. 2024Adam with model exponential moving average is effective for nonconvex optimizationcited by 23
  2. 2022Language Control Diffusion: Efficiently Scaling through Space, Time, and Taskscited by 20
  3. 2021Towards Learning to Play Piano with Dexterous Hands and Touchcited by 17
  4. 2022Learning Reward Functions for Robotic Manipulation by Observing Humanscited by 17
  5. 2025Diffusion Trajectory-guided Policy for Long-horizon Robot Manipulationcited by 15
  6. 2024ViSaRL: Visual Reinforcement Learning Guided by Human Saliencycited by 12
  7. 2023Manipulate by Seeing: Creating Manipulation Controllers from Pre-Trained Representationscited by 10
  8. 2024Adapt2Reward: Adapting Video-Language Models to Generalizable Robotic Rewards via Failure Promptscited by 10
  9. 2024Switch EMA: A Free Lunch for Better Flatness and Sharpnesscited by 9
  10. 2024PianoMime: Learning a Generalist, Dexterous Piano Player from Internet Demonstrationscited by 9
  11. 2024Batch Active Learning of Reward Functions from Human Preferencescited by 8
  12. 2021Learning Robotic Manipulation Tasks via Task Progress based Gaussian Reward and Loss Adjusted Explorationcited by 2

Built on by

Cited by year

Citing papers per year: 2025: 2.

2 papers.

  1. 2025Learning to Play Piano in the Real Worldcited by 2
  2. 2025FORGE-Tree: Diffusion-Forcing Tree Search for Long-Horizon Robot Manipulationcited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2026Expressive Robotic Pianist: Mastering Complex Piano Repertoire with Graph-Mimic and Musical Dynamics ↗
  2. 2026HandelBot: Real-World Piano Playing via Fast Adaptation of Dexterous Robot Policies
  3. 2024RP1M: A Large-Scale Motion Dataset for Piano Playing with Bi-Manual Dexterous Robot Hands
  4. 2025Dexterous Robotic Piano Playing at Scale
  5. 2024DExter: Learning and Controlling Performance Expression with Diffusion Models
  6. 2026MUSIC: Learning Muscle-Driven Dexterous Hand Control
  7. 2025Robot Drummer: Learning Rhythmic Skills for Humanoid Drumming
  8. 2026DexDrummer: In-Hand, Contact-Rich, and Long-Horizon Dexterous Robot Drumming ↗