ManiDP: Manipulability-Aware Diffusion Policy for Posture-Dependent Bimanual Manipulation

2025-10-27

AI Summary

Existing dual-arm manipulation approaches neglect task characteristics that are inherently dependent on robot pose, hindering adaptive satisfaction of dynamic force and velocity requirements in dexterous manipulation. To address this, we propose a manipulability-aware diffusion policy: the first to embed pose-dependent robotic manipulability priors into a diffusion model, employing Riemannian-space probabilistic encoding to represent pose features and explicitly guiding dual-arm trajectory generation under task constraints during conditional sampling. Our method unifies imitation learning, manipulability analysis, and geometry-aware probabilistic modeling. Evaluated on six real-world dual-arm tasks, it achieves an average success rate improvement of 39.33% and a 0.45 increase in task compatibility over baselines—demonstrating superior pose adaptivity, dexterity, and human-like skill generation.

Technology Category

• Intelligent Robots: Manipulation
• Computer Vision: Diffusion Models for Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Recent work has demonstrated the potential of diffusion models in robot bimanual skill learning. However, existing methods ignore the learning of posture-dependent task features, which are crucial for adapting dual-arm configurations to meet specific force and velocity requirements in dexterous bimanual manipulation. To address this limitation, we propose Manipulability-Aware Diffusion Policy (ManiDP), a novel imitation learning method that not only generates plausible bimanual trajectories, but also optimizes dual-arm configurations to better satisfy posture-dependent task requirements. ManiDP achieves this by extracting bimanual manipulability from expert demonstrations and encoding the encapsulated posture features using Riemannian-based probabilistic models. These encoded posture features are then incorporated into a conditional diffusion process to guide the generation of task-compatible bimanual motion sequences. We evaluate ManiDP on six real-world bimanual tasks, where the experimental results demonstrate a 39.33$%$ increase in average manipulation success rate and a 0.45 improvement in task compatibility compared to baseline methods. This work highlights the importance of integrating posture-relevant robotic priors into bimanual skill diffusion to enable human-like adaptability and dexterity.

Problem

• Addresses posture-dependent task feature learning in bimanual manipulation
• Optimizes dual-arm configurations for force and velocity requirements
• Incorporates manipulability features into diffusion policy for task compatibility

Innovation

• Uses manipulability features from expert demonstrations
• Encodes posture features with Riemannian probabilistic models
• Guides motion generation via conditional diffusion process