The 80s television action hero McGiver could not slip away with the help of a disposable lighter, Penkenife and two tabs of Alka Seltzer.  His unconventional application of scientific principles and out-of-the-box thinking made him such a formidable opponent week after week.  Now, a team of researchers from Georgia Tech's Research Research Lab are working to establish the same survival skills in robots.
Georgia Tech Researchers Teach Robots to Become Mechanical McGivers

The RAIL team, led by PhD student Lakshmi Nair, focused their research on teaching robots to fabricate tools - hammers, screw drawers, ladles and whatever else is out of the material.  "So if a robot needs to solve a task that uses a hammer," Nair explained to Engdg.  "They can combine a stick and a stone, for example, to be able to make a hammer.

But it's just asking the robot to show a picture of a hammer and make it into something that looks like it.  "It's situation-specific," Nair said.  "So looking at a particular situation, if it wants to hammer something, it figures out which items to put together. So you're not giving it a specific example of a hammer, just state that  What's the situation. 
Georgia Tech Researchers Teach Robots

Once the robot knows what it needs to do, it will evaluate the contents of its workspace based on its size and how they can be interconnected.  Taking advantage of supervised machine learning, the RAIL team taught the system to match objects by their relative shapes and perceived functions.

Essentially, robots are taught to match form to function so that it learns that things like bowl flatness enable it to hold liquids, for example," Nair said.  The system is given examples labeled everyday objects so that when it is presented with a new set of objects that it has not seen before, it can use what it has learned to reason about previously unknown objects.  Has learned  A joint of cups and tongs becomes a rectangular flattened screwdriver with the same tongs and a coin to make the ladles, while piercing a rectangular foam block with a poker makes a DIY squeeze.

Although the robot uses a handheld spectrometer to determine whether an item is pierceable or not, it cannot effectively see what it is looking at, which has led to many hammers from the foam block  Is coined  "Said right, it just looks at size as one of the main reasons for using what parts to build the equipment," Bair explained.  "Some of the work going on in our future looks to include material tips so that it is constructed of hammers or more robust material that can be used in a real application."

Nair's team is not alone in its robot tool building research.  In fact, a team at Tufts University suggested the McGiver test in 2017 as a means of evaluating the resourcefulness and creativity of a robot - a practical alternative to the Turing test, which scores the robot's dysfunction.

Tuff's team wrote, "The proposed evaluation framework," based on the idea of ​​a McGiver-esque creativity, to answer the question of whether embodied machines can generate, execute, and create real-world imaginable.  Can identify strategies to identify and solve problems.  
Georgia Tech Researchers

But is it shrewdness or resourcefulness?  In the TV show, McGiver's gadgets and contraceptives were so effective because his plans exploited some underlying scientific or physics theory.  Professor Nathan Michael, director of RISLab at Carnegie Mellon University, argues that these systems are performing a creative task similar to the character of Richard Dean Anderson.

A lot of the problem-solving challenges that arise when systems are entangled in executive-level task planning share a lot of similarities with this idea of ​​building tools," he told Angadze.  "In fact, the underlying algorithmic framework employed for the idea of ​​problem-solving is basically trying to optimize or find a solution within a particular set of constraints subject to a particular situation.

The way these problems are handled depends on the chosen methodology, which in turn "depends largely on the nature of the problem".  "What we're talking about here is like a big problem of logic. The question of shrewdness or resourcefulness really comes down to, in this context, the ability to determine the amount of systems that can do it or not.  Those resources, it cannot find and find a way to solve that particular problem through the application of those resources and leverage its potential. "

There are many places in the real world where using such systems Nair said, "Any application that includes repair and repair".  In addition to helping repair the light house, Nair speculates that this technology will make it into space before too long.  "You have space exploration where you could potentially send these robots in advance and then use the available resources to build habitats before humans could get there," she continued.  However, the system still faces many technical challenges - whether they be clunky manipulator weapons or cameras that cannot reliably detect metal items - before we see them in space.