Interactive Identification of Granular Materials using Force Measurements

2024-03-26

arXiv.org

AI Summary

Granular material identification is critical for robotic applications such as cooking and excavation, demanding vision-free, real-time discrimination—yet existing approaches rely on shaking materials within enclosed containers, limiting practical applicability. This paper introduces the first active dynamic interaction-based paradigm for granular material identification: it employs only force-torque sensing during controlled, contact-based physical exploration to extract interpretable time-frequency mechanical features, which are classified using a lightweight machine learning model. Key contributions include: (1) the first interactive mechanical perception framework for granular material identification; (2) the first publicly available, multi-class (11-category) benchmark dataset of force-sensor measurements for granular materials; and (3) state-of-the-art performance achieving 96.2% mean classification accuracy in real-world robotic settings. Both code and dataset are open-sourced to facilitate rapid integration into robotic manipulation pipelines.

Technology Category

• Intelligent Robots: Manipulation
• Machine Learning: Active Learning
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

The ability to identify granular materials facilitates the emergence of various new applications in robotics, ranging from cooking at home to truck loading at mining sites. However, granular material identification remains a challenging and underexplored area. In this work, we present a novel interactive material identification framework that enables robots to identify a wide range of granular materials using only a force-torque sensor for perception. Our framework, comprising interactive exploration, feature extraction, and classification stages, prioritizes simplicity and transparency for seamless integration into various manipulation pipelines. We evaluate the proposed approach through extensive experiments with a real-world dataset comprising 11 granular materials, which we also make publicly available. Additionally, we conducted a comprehensive qualitative analysis of the dataset to offer deeper insights into its nature, aiding future development. Our results show that the proposed method is capable of accurately identifying a wide range of granular materials solely relying on force measurements obtained from direct interaction with the materials. Code and dataset are available at: https://irobotics.aalto.fi/indentify_granular/.

Problem

• Identifying granular materials using direct force-torque measurements during robot interaction
• Developing feature space combining time-domain dynamics and frequency spectrum of force signals
• Creating public dataset and framework for accurate granular material identification

Innovation

• Uses direct interaction with granular materials
• Relies solely on force-torque measurements for identification
• Combines time-domain and frequency features in representation