[Paper Review] Tool Macgyvering: A Novel Framework for Combining Tool Substitution and Construction

Lakshmi Nair, Nithin Shrivatsav Srikanth+1 more|arXiv (Cornell University)|Aug 24, 2020

Manufacturing Process and Optimization38 references4 citations

TL;DR

This paper introduces a novel Tool Macgyvering framework that unifies tool substitution and construction via arbitration, using shape, material, and attachment reasoning to autonomously create or substitute tools. The system achieves 96.67% accuracy in constructing functional tools and 83.33% accuracy in selecting between substitution and construction, significantly improving over prior methods through integrated material reasoning and direct comparison-based arbitration.

ABSTRACT

Macgyvering refers to solving problems inventively by using whatever objects are available at hand. Tool Macgyvering is a subset of macgyvering tasks involving a missing tool that is either substituted (tool substitution) or constructed (tool construction), from available objects. In this paper, we introduce a novel Tool Macgyvering framework that combines tool substitution and construction using arbitration that decides between the two options to output a final macgyvering solution. Our tool construction approach reasons about the shape, material, and different ways of attaching objects to construct a desired tool. We further develop value functions that enable the robot to effectively arbitrate between substitution and construction. Our results show that our tool construction approach is able to successfully construct working tools with an accuracy of 96.67%, and our arbitration strategy successfully chooses between substitution and construction with an accuracy of 83.33%.

Motivation & Objective

• To enhance robot adaptability by enabling autonomous improvisation of tools using available objects.
• To address the limitations of prior tool construction methods that ignored material properties, leading to non-functional tools.
• To develop a unified framework that arbitrates between tool substitution and construction for optimal macgyvering solutions.
• To improve performance and reliability of robotic tool improvisation in unstructured environments.
• To validate the framework through autonomous tool construction on a 7-DOF robotic arm and human-evaluated arbitration decisions.

Proposed method

• The framework uses independent shape, material, and attachment scoring for tool construction, enabling the robot to reason about object compatibility and functionality.
• Material reasoning is integrated into tool construction by evaluating object properties such as rigidity and durability to ensure functional tool creation.
• Tool substitution employs joint shape and material scoring to identify the best available object as a substitute for a missing tool.
• Arbitration between substitution and construction is performed via direct comparison of composite scores, outperforming rule-based or indirect comparison strategies.
• The system leverages RGB-D sensors to capture partial point clouds of objects and uses geometric and physical reasoning to evaluate potential tool configurations.
• The framework is evaluated on a 7-DOF robotic arm using a diverse set of objects with varied shapes and materials, testing six different tool types.

Experimental results

Research questions

• RQ1 Can material reasoning significantly improve the success rate of robotic tool construction compared to shape-only approaches?
• RQ2 How effective is direct comparison of composite scores in arbitrating between tool substitution and construction?
• RQ3 What is the impact of integrating shape, material, and attachment reasoning on the diversity and functionality of constructed tools?
• RQ4 How does the performance of the arbitration strategy compare to rule-based or indirect comparison methods in real-world macgyvering scenarios?
• RQ5 To what extent does the system reduce the configuration space explored while maintaining high construction accuracy?

Key findings

• The tool construction approach achieved a 96.67% success rate in creating functional tools, a significant improvement over prior methods that achieved only 27% and 60% completion rates.
• The integration of material reasoning reduced the creation of non-functional tools (e.g., foam hammers), increasing the reliability and practicality of constructed tools.
• The arbitration strategy based on direct comparison of scores correctly selected the best macgyvering solution (substitution or construction) in 83.33% of test cases, outperforming rule-based and substitution-based baselines.
• The system explored only 5.72% of the total possible configuration space, demonstrating efficient search within large combinatorial spaces.
• The best-performing framework design combined independent shape, material, and attachment scoring for construction, joint shape and material scoring for substitution, and direct comparison for arbitration.
• Human evaluations confirmed that the direct comparison arbitration strategy aligned most closely with ground truth preferences, validating its effectiveness in real-world decision-making.