This morning I was alerted to the following research paper by someone claiming that scientists have created some small, ant-like robots that exhibit “exbodied” intelligence:

Robotectonics: Emergent phototactic aggregation-disaggregation in swarms [Source]

Interesting Engineering described the study this way:

New robot ants work like real insects to build and dismantle on their own

It is based on “exbodied intelligence,” where coordination arises from interacting with the environment rather than complex internal programming.

. . .

I don't understand the electrical engineering aspects of the experimental study, but I do understand how living things work and how they behave. These so-called “robot ants” ('RAnts') are not even close to demonstrating “exbodied intelligence,” nor any sort of intelligence at all.

I encourage you to take a look at the videos for yourself. They're a series of five mp4 files — they are titled SIVid_1.mp4 to SIVid_5.mp4 — in the Supplemental Material section at the end of the paper.

What I saw in the highly curated videos was a series of experiments showing rudimentary and bumbling machines that are a far cry from the natural intelligence of ants.

These rudimentary and bumbling machines are not even close

to demonstrating intelligence

. . .

Here is what I saw in each of the five videos:

Video 1. The robot circles aimlessly within its photonic corral (pool of light) when the radius of the lighted area is much larger than the robot (this happens with all radii > 10 mm).

At the larger radii (20–35 mm), the robot shows no affinity for the bright pool of light in the test area. The robot cycles in and out of the brightly lit area, no longer forming a circle (more of an egg shape, complete with a narrow end and a broad end), and not even trying to follow the fairly clear margin of the bright light.

Perhaps this light effect is simply an artefact of the videography. But as it is presented to us, this experiment simply shows the robot making a shape that is more reminiscent of a chicken's egg than a purposeful movement directed by a strong affinity for the signalling factor, which in this instance is light.

The robot circles aimlessly, showing no purposeful movement

or strong affinity for the signalling factor

. . .

Video 2. When multiple robots are placed in a large, square test area, they separately circle within a fairly narrow and self-proscribed radius. They show no affinity for, or communication with, the other robots. They are neither self-organizing nor 'swarming'.

They circle relentlessly, although aimlessly, like the repetitive and purposeless (stereotypic) behavior of captive animals who are bored or stressed.

The robots are neither self-organizing nor 'swarming'

. . .

Video 3. The three robots accidentally bump into each other — first just two, and much later the third — and they inadvertently end up moving together in a roughly circular pattern that is dictated by their geometry and programmed pattern of movement.

The first two robots accidentally bump into each other, their long sides eventually lining up as the two circle independently, effectively tethering the two robots in their programmed pattern of movement. Their tethered movement forms an elliptical pattern, as they are not moving together in any coordinated way.

The third robot accidentally T-bones the pair and is then drawn along with the first two by their geometry and programmed pattern of movement.

The start of the video shows aimless paths of movement for all three robots before the collisions, which appear to occur entirely by chance.

The robots collide by chance and become tethered

by their geometry and programmed pattern of movement

. . .

Video 4. The robots randomly bump into each other and the sides of the test area. Their blundering around randomly disassembles anything even remotely like a collection of objects. It is reminiscent of the Keystone Cops.

Their magnetic grasping of objects appears purely accidental or incidental (e.g., when they blunder into the wall, they magnetically pick up an object), and the human hand that repeatedly appears and adds more objects in the area in which one or more robots are “working” gives lie to the notion that these robots are acting in any way deliberatively or purposefully, or to any sort of intended design or architecture.

The robots repeatedly bump into, and bounce off, the objects they're supposedly organizing into architectural structures, destroying the structures (really just a random collection of objects) in the process.

The robots blunder around, randomly disassembling

anything remotely like a collection of objects

. . .

Video 5. The robots aimlessly deconstruct a human-built wall of objects in a way that incoherently builds it up further in adjacent areas and ends up boxing them in.

Whereas the light is supposed to guide them to the area of wall to disassemble, the robots spend a good deal of time aimlessly wandering around on the other side of the test area, and in unlit areas.

Overall, they do not appear to be guided by, or responsive to, light at all. Although, one gets so confused that it goes manic in the lower-right corner, where there is a pool of light. That corner is just about the furthest point from the area of wall some of the other robots are supposedly disassembling.

The robots aimlessly and incoherently deconstruct a manmade wall,

and they do not appear to be responsive to light at all

. . .

In summary, the robots fail to make any sort of structure that is in any way consistent in size, shape, complexity, or apparent purpose. Nor are they able to work together to deconstruct a man-made structure.

These robots don't even come close to demonstrating any sort of organization, nor any ability to learn and respond to their environment as the authors claim. They don't build anything — and they can't even manage that on their own. A human hand keeps appearing and moving around the objects the robots are supposed to be using to assemble and disassemble their ant cities.

And really, what is the point of a robot that could be led astray by an errant  light, like a cat chasing a laser pointer?

What is the point of a robot that can be led astray

like a cat chasing a laser pointer?