Task Allocation for a robotic swarm based on an Adaptive Response Threshold Model

Adaptive foraging for simulated and real robotic swarms: the dynamical response threshold approach

“…Preliminary simulation results obtained with ARTM were presented in Castello et al (2013) and Castello et al (2014). In addition to the more extensive results included in this paper, we demonstrate that ARTM can achieve adaptive division of labour in a small size robot swarm in an efficacious and robust way.…”

“…The benefits of using dynamical threshold models (such as ARTM), in terms of increase of the survivability and adaptability of swarm systems in changing environments, was previously shown only by computer simulations in Castello et al (2013) and Castello et al (2014). This paper aims at confirming those simulation results in a new set of experiments conducted using real robots.…”

“…The Adaptive Response Threshold Model (ARTM), first described in Castello et al (2013), is an extension of the classical Fixed Response Threshold Model (FRTM), in which the response threshold (θ) is calculated dynamically instead of remaining fixed. ARTM, like other response threshold methods, strongly relies on the concept of stimulus:…”

“…The major difference with previous fixed response threshold models is the Discrete Attractor Selection Model (DASM), represented by a rectangular box with dashed boundaries. DASM is the algorithm first introduced in Castello et al (2013), which dynamically calculates the θ threshold according to different environmental conditions. Once the process of updating θ is over, robots are able to calculate their response curve (defined in equation (2)) according to the new conditions and decide whether to begin the foraging process or wait, in standby mode, near the nest.…”

“…In foraging scenarios, θ has a great influence on the robot's behaviour, because it determines when the robot goes foraging (i.e., to expend energy looking for food within the environment), or rest (and conserve energy by staying in the nest) in case its work is not needed at that specific moment (Labella et al, 2006;. On one hand, prior work in the field has shown that a high value of θ can decrease interference among robots and therefore increase the system's performance (Zhang and Zeng, 2012), while on the other hand lower values of θ can beneficially affect the adaptation in response to abrupt changes in the availability of food in the environment (Castello et al, 2013). Tuning and optimising θ according to different environmental conditions in order to make a robot swarm adapt to dynamical situations is therefore not straightforward.…”