Instead of building increasingly complex robotic systems to carry out even more complex tasks, Harvard University researchers have developed a fleet of ant-inspired simple robots called “RAnts” that can use environmental cues and minimal physical rules to carry out complex activities, including spontaneously organizing to build or dismantle complex structures without detailed instructions or a team leader.
The team behind the robotic ants said the work shows how a decentralized swarm of insects or robots can coordinate activities to complete tasks without a detailed plan or centralized control, using simple cues and rules.
Along with offering potentially invaluable insights into how both biological systems and future autonomous robotic systems work together and helping refine experimental models used to study animal behavior, the team suggested that simplified bio-inspired robotic swarms could be designed to carry out complex tasks in hazardous or hard-to-reach environments that pose challenges for human technicians, including applications in planetary exploration.
Rants Replace Pheromones with ‘Photormones’ to Coordinate Activities
Many animals work together to achieve a common goal, but ants have continued to fascinate scientists for their ability to build massive, intricate, climate-controlled structures despite their comparatively simple neurological system. As the Harvard team noted, ants also complete these complex structures without a blueprint or “worksite foreman” directing the process.
Instead of using a complex, pre-planned approach, insects like termites and ants that work together to build entire structures employ a technique called stigmergy. According to the Harvard team, stigmergy is a process in which individual members “modify” their environments and respond to those modifications. For ants, this process involves the emission of pheromones that other ants detect and respond to.
To replicate stigmergy for Rants, team leader Professor L. Mahadevan, whose lab has studied social insect communities for several years, also showed that robotic ants could achieve similar coordinated tasks, essentially replacing chemical pheromones with visible, light-based “photormones.” According to the professor, photormones are light fields that operate as “digital stand-ins” for insect pheromone fields.
In practical terms, each individual RAnt is designed to sense gradients in the photormone field the same way an ant would sense gradients in the pheromone field and leave behind similar signals as it moves. The team said this process creates a “feedback loop between robots and their environment,” enabling coordinated activities across the entire swarm.
The Three Simple Rules That Trigger ‘Sophisticated’ Behaviors
Eerily reminiscent of Isaac Asimov’s “Three Laws of Robotics” from the ‘I, Robot‘ series, the Harvard robotics team programmed the RAnts with three basic rules. The first rule is to always follow the gradients in the photormone signal. The second rule is to pick up and transport building blocks in response to the photormone cues, and the third rule is to deposit materials when “signal thresholds” are met.
Although they sound basic, the team noted that these simple rules “nonetheless trigger sophisticated behaviors.” For example, they noted that robots programmed with these three rules “spontaneously cluster to form nucleation sites,” where the first phases of individual structures begin to emerge.
According to the researchers, these collective nucleation sites arise through a mechanism termed “trapping instability.” When robots enter this area, they are temporarily confined by the photormone signals they are actively generating.
Over time, more Rants converge on the nucleation sites, accelerating the construction process. The team says that this seemingly simple set of rules results in “organized aggregates of building material.”
After some experimentation, the researchers discovered that the robot swarm’s behavior could be ‘tuned’ by adjusting two parameters. The first parameter, called “cooperation strength,” governs how strongly a robot follows the signal gradient. The second parameter, called “deposition rate,” dictates whether a robot deposits or removes material. Depending on the setting of these two parameters, the team could switch the RAnt swarm between structure construction and demolition.
Potential SOlutions in Hazardous Environment COnstruction and Planetary Exploration
When discussing the implications of their RAnt swarm capable of creating complex structures, Professor Mahadevan and colleagues highlighted the refreshingly basic nature of their approach.
“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,” he explained.
The Lola England de Valpine Professor of Applied Mathematics, Organismic and Evolutionary Biology, and Physics at SEAS and FAS also noted that his team’s work introduced the concept of ‘exbodied intelligence,’ “where collective cognition arises not solely from individual agents, but from their ongoing interaction with an evolving environment.”
Although the RAnts are still confined to the lab setting, the research teams’ findings could lead to applications in autonomous construction in hazardous environments, planetary exploration, and experimental models for studying animal behavior.
The study “Robotectonics: Emergent Phototactic Aggregation-Disaggregation in Swarms” was published in PRX Life.
Christopher Plain is a Science Fiction and Fantasy novelist and Head Science Writer at The Debrief. Follow and connect with him on X, learn about his books at plainfiction.com, or email him directly at christopher@thedebrief.org.