Tool Macgyvering: Tool Construction Using Geometric Reasoning

Abstract

MacGyvering is defined as creating or repairing something in an inventive or improvised way by utilizing objects that are available at hand. In this paper, we explore a subset of Macgyvering problems involving tool construction, i.e., creating tools from parts available in the environment. We formalize the overall problem domain of tool Macgyvering, introducing three levels of complexity for tool construction and substitution problems, and presenting a novel computational framework aimed at solving one level of the tool Macgyvering problem, specifically contributing a novel algorithm for tool construction based on geometric reasoning. We validate our approach by constructing three tools using a 7-DOF robot arm.

PIQA: Reasoning about Physical Commonsense in Natural Language

“…Robotics. Learning from interaction and intuitive physics (Agrawal et al 2016) can also be encoded as priors when exploring the world (Byravan et al 2018) and internal models of physics, shape, and material strength enable advances in tool usage (Toussaint et al 2018) or construction (Nair, Balloch, and Chernova 2019). Key to our research aims in this work is helping to build language tools which capture enough physical knowledge to speed up the bootstrapping of robotic-language applications.…”

Force-and-Motion Constrained Planning for Tool Use

“…Brown and Sammut presented an architecture for learning how to use objects as tools within a problem-solving context (Brown and Sammut, 2007) and Wicaksono and Sammut later extended this onto a real robot platform (Wicaksono and Sammut, 2016). Nair et al demonstrated a system that constructed conventional tools from objects available in environment by reasoning over part shapes and potential ways to combine those shapes (Nair et al, 2019). Each of the example tool-use tasks in this thesis are quasistatic.…”

Feature Guided Search for Creative Problem Solving Through Tool Construction

“…Prior work by Nair et al introduced a novel computational framework for performing tool construction, in which the approach takes an input action, e.g., “flip,” in order to output a ranking of different object combinations for constructing a tool that can perform the specified action, e.g., constructing a spatula (Nair et al, 2019a , b ). For performing the ranking, the approach scored object combinations based on the shape and material properties of the objects, and whether the objects could be attached appropriately to construct the desired tool.…”

“…Hence, our work relaxes a key assumption of the prior work that requires the input action to be specified. Our approach directly uses the score computation methodology described in prior work (Nair et al, 2019a , b ), but combines it with planning heuristics to integrate tool construction within a task planning framework.…”