BagIt! An Adaptive Dual-Arm Manipulation of Fabric Bags for Object Bagging

2025-09-11

AI Summary

Manipulating deformable fabric bags for automated packing using dual robotic arms in industrial settings remains challenging due to complex, unpredictable bag deformation, unknown physical properties, and difficulty in modeling. Method: This paper proposes a closed-loop control framework based on an adaptive Structure of Interest (SOI). It integrates Gaussian Mixture Model (GMM)-driven SOI state estimation, constraint-aware Bidirectional Rapidly-exploring Random Tree (CBiRRT) motion planning, Model Predictive Control (MPC) for coordinated dual-arm execution, and real-time visual feedback for dynamic adjustment. Contribution/Results: The key innovation lies in the SOI strategy, which requires no prior knowledge of object physical parameters; instead, it leverages data-driven state representation and generative optimization to enhance robustness and cross-object generalization in model-free scenarios. Experiments demonstrate high accuracy and strong adaptability across diverse packing tasks, advancing the practical deployment of deformable object manipulation in industry.

Technology Category

• Intelligent Robots: Manipulation
• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Bagging tasks, commonly found in industrial scenarios, are challenging considering deformable bags' complicated and unpredictable nature. This paper presents an automated bagging system from the proposed adaptive Structure-of-Interest (SOI) manipulation strategy for dual robot arms. The system dynamically adjusts its actions based on real-time visual feedback, removing the need for pre-existing knowledge of bag properties. Our framework incorporates Gaussian Mixture Models (GMM) for estimating SOI states, optimization techniques for SOI generation, motion planning via Constrained Bidirectional Rapidly-exploring Random Tree (CBiRRT), and dual-arm coordination using Model Predictive Control (MPC). Extensive experiments validate the capability of our system to perform precise and robust bagging across various objects, showcasing its adaptability. This work offers a new solution for robotic deformable object manipulation (DOM), particularly in automated bagging tasks. Video of this work is available at https://youtu.be/6JWjCOeTGiQ.

Problem

• Automating dual-arm manipulation of deformable fabric bags
• Adaptive bagging without prior bag property knowledge
• Real-time visual feedback for dynamic action adjustment

Innovation

• Adaptive SOI strategy for dual arms
• Real-time visual feedback without prior knowledge
• GMM, CBiRRT, and MPC integration