In the EU-FET project subCULTron we aim for a novel way of long-term and large-scale environmental monitoring of endangered ecosystems and habitats. We chose the lagoon of Venice as a benchmark site of significant impact. The goal was not only to let some robots run for a very long time in out-of-the lab conditions, what would already be a challenge in itself. Going significantly beyond this challenge, we let them act like one big unit - as a collective. This collective, you might also call it "a swarm", should operate significantly longer than the period of time that an individual life cycle (the period between one charging of a robot to its next necessary recharging) of a single robot lasts. Thus, the swarm needs to have the ability of autonomous (re-)charging and energy transfer from one unit to the other. To allow an efficient and effective operation, we created three different robot types, each one designed for pursuing specific objectives under specific constraints, thus these robot "species" are very different when compared to each other: These difference are in size, in power, in agility and in energy cycle durations. Such a high degree of specialization leads to a very heterogeneous robot swarm, what poses another set of challenges. The specialization of subCULTron's robotic "species" was achieved by looking for analog functionalities achieved by living organisms in nature and by drawing bio-inspiration from these sources of inspiration. This led to design of three unique bio-inspired and biomimetic robot "species", as they are shown and briefly described below. The interoperability of these 125 autonomous robots allows not only to monitor the environment over long periods of time and on large areal scales but also to spread and process these datasets collectively across the swarm system. Based on this shared and distributed information processing the subCULTron robot swarm can adapt its mode of operation, thus it learns during its runtime. This learning extends beyond the individual life cycle (energy cycle) of each single swarm member, thus we call it to be a form of "cultural learning" in a swarm robotic context, an aspect that defines the "CULT" in the name of the subCULTron project.

The aMussel robots are the key agents for our environmental monitoring activities. They need to coordinate their activities, e.g. where to stay and where to reposition via drifting or via an active aPad transport. Here we see a group of 3 aMussels that are deployed one after the other, establish contact via blinking signals with each other and, after some time of staying at the bottom of the water body, decide when to go back to the surface in order to be moved to another location by an aPad.