A new study has found that simple ant-inspired robots can build clusters and dig through material without a boss, a map, or a step-by-step plan.
The result points toward machine teams that can keep working in messy places where central commands break down.
Inside a small test arena, RAnts – small wheeled robots modeled after ants – carried plastic blocks while reacting only to light below them.
The research was led by Professor L. Mahadevan at Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS).
By tracking the swarm’s light traces, Mahadevan’s team showed that local cues could drive construction and dismantling.
The researchers used those cues to replace central commands with feedback.
Ant behavior shapes design
Ants and termites often coordinate by changing nearby spaces, including termite mounds, tall nest structures that can ventilate colonies with daily temperature swings.
Biologists call this stigmergy, indirect coordination through changes left in a shared space, and it turns small actions into shared guidance.
In insects, pheromones – chemicals animals leave as signals – can mark paths that other workers follow or reinforce.
For the robots, projected light served the same job, so a trail could guide later movement without speech or radio.
Signals guide the swarm
The robots followed photormones – projected light trails that stand in for pheromones – rather than scent.
Each robot sensed brighter or dimmer light under two sensors, then turned toward stronger traces as it moved.
Every pass added more light to the floor, and the signal faded after the robots moved away.
Fresh trails pulled more robots together, while fading trails kept old activity from ruling the arena forever.
Clusters create work areas
Repeated light trails created nucleation sites – starting spots where building is first concentrated – when several robots circled the same area.
That feedback became a trapping instability, a loop that temporarily confines robots, because their own signal keeps attracting them back.
Experiments tuned the robots so a trap usually needed five or more robots, which kept a lone machine from dominating.
“Our new study shows how simple, local rules can lead to the emergence of complex task completion that is self-organized and thus robust and adaptive,” said Mahadevan.
Simple rules shift tasks
Once traps formed, the swarm’s behavior changed when researchers adjusted only two settings in the control rules.
Cooperation strength, or how strongly robots follow brighter parts of a trail, pulled them toward common work zones when set high.
Deposition rate – how quickly robots drop or remove material – decided whether those zones gained blocks or lost them.
With both settings in the right range, the same simple rules produced different jobs instead of one fixed routine.
Building and digging emerge
Changing the deposition setting flipped the task from construction to excavation, without adding a supervisor.
When the setting favored dropping material, robots gathered blocks into organized piles where the light signal grew.
During removal, the machines carried material away and opened paths through an existing structure instead.
Earlier ant and robot excavation work showed how simple animals and machines can dig together, but this system handled both directions.
Theory explains the swarm
Beyond the arena, the SEAS team wrote a continuum model, equations that describe the group as flowing densities.
Instead of tracking each wheel turn, the model followed robot density, light signals, and material through space.
That choice linked individual rules to group patterns, including scattered piles, dense clusters, and excavation fronts.
A phase space, a map of possible group behaviors, let researchers choose building or digging without rewriting the robots.
Practical value emerges
Central control can fail when machines enter collapsed buildings, disaster zones, or distant planets with slow communication.
Local rules offer another route, because each robot only needs nearby cues and a way to move material.
The research team described this as exbodied intelligence – coordination built through workers and surroundings – because the workspace records useful recent activity.
Practical uses could include hazardous construction or planetary exploration, yet real swarms still need stronger hardware and safety rules.
Gaps to guide research
Current experiments used simple blocks, projected light, and a controlled floor – not an open construction site.
Because the arena stayed limited, the robots did not choose which final shape worked best for a real task.
Future systems may need outcome selection – rules that favor useful structures over merely possible ones.
Even so, the result gives engineers a cleaner way to test cooperation before adding expensive machines.
Natural teamwork emerges
Across robots and insects, the main finding is evident: groups can organize when actions leave traces that others can read.
Such evidence may help scientists compare insects and machines while keeping limits clear: simple rules do not remove the hard work of design.
The study is published in the journal PRX Life.
—–
Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.
Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.
—–