Testing the limits of pheromone stigmergy in high-density robot swarms

Abstract

Area coverage and collective exploration are key challenges for swarm robotics. Previous research in this field has drawn inspiration from ant colonies, with real, or more commonly virtual, pheromones deposited into a shared environment to coordinate behaviour through stigmergy. Repellent pheromones can facilitate rapid dispersal of robotic agents, yet this has been demonstrated only for relatively small swarm sizes (N < 30). Here, we report findings from swarms of real robots (Kilobots) an order of magnitude larger (N > 300) and from realistic simulation experiments up to N = 400. We identify limitations to stigmergy in a spatially constrained, high-density environment—a free but bounded two-dimensional workspace—using repellent binary pheromone. At larger N and higher densities, a simple stigmergic avoidance algorithm becomes first no better, then inferior to, the area coverage of non-interacting random walkers. Thus, the assumption of robustness and scalability for such approaches may need to be re-examined when they are working at a high density caused by ever-increasing swarm sizes. Instead, subcellular biology, and diffusive processes, may prove a better source of inspiration at large N in high agent density environments.

Spider webs inspiring soft robotics

“…In computer engineering, the concept of stigmergy serves as inspiration for applications in wide-ranging areas of animal and human behaviour wherever an action configures the following action by the intermediary of the trace it leaves [ 49 ]. Examples for the phenomenon range from ant nest construction [ 48 , 50 ] and biofilm self-organization [ 51 ] to robot swarm behaviour [ 52 ] and satellite clusters [ 53 ] as well as the development of free open source computer programs [ 54 ].…”

HuGoS: A Multi-user Virtual Environment for Studying Human–Human Swarm Intelligence

“…In order to study decision making based on external information, of the type studied in human groups in platforms such as UNUM [40], landmarks in the environment could be labelled with each option, and participants could use their avatar positions in relation to the landmarks to indicate their opinions. The third class involves the agents making changes to the environment, such as in stigmergic communication (e.g., [12]) or in the performance of a task such as collective construction (e.g., [51]). In HuGoS, this requires game objects that can be manipulated or modified by avatars-for instance, immovable environment features with modifiable properties such as color, or movable objects such as construction blocks.…”

Phormica: Photochromic Pheromone Release and Detection System for Stigmergic Coordination in Robot Swarms

“…A number of researchers have proposed different approaches in which an external infrastructure is used to partially or completely store the stigmergic information. For instance, the movement of robots is tracked using an overhead camera and the pheromone trails are then either projected on the ground (Hamann et al, 2007 ; Garnier et al, 2013 ; Hunt et al, 2019 ; Talamali et al, 2020 ) or displayed on LCD screens that serve as the floor on which the robots move (Arvin et al, 2015 ; Na et al, 2019 , 2020 ). The robots can detect the projected pheromone trails locally using their sensors, for then acting accordingly.…”

“…One could argue that this method is not exactly using indirect communication to achieve stigmergy since robots rely on the presence of their peers. Other research works exist in which robots release artificial pheromone trails: trails of alcohol detected by chemical sensors (Russell, 1999 ; Fujisawa et al, 2014 ); trails of colored ink on paper (Svennebring and Koenig, 2004 ); trails of heat (Russell, 1997 ); virtual trails using an external infrastructure to record the positions of robots and project trails on the ground, or display them on a large LCD screen that acts as the floor (Hunt et al, 2019 ; Na et al, 2019 ); RFID trails where tags are installed in the environment for the robots to interact with (Khaliq et al, 2014 ; Khaliq and Saffiotti, 2015 ); and trails of phosphorescent paint (Mayet et al, 2010 ). Despite the importance of these approaches in the literature, we find that they have several limitations.…”