Browsing by Author "Hu, Huosheng"

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    A model for using self-organized agents to visually map environmental profiles
    (Elsevier, 2014-06-09) Oyekan, John; Gu, Dongbing; Hu, Huosheng
    In this work, we investigate the possibility of using inspiration from the self-organizing property of organisms in nature for providing visual representation of an invisible pollutant profile. We present a novel mathematical model of the bacterium and use it to find pollutants in the environment. This model has the capability of exploring the environment to search for sparsely distributed pollutants or food sources and then subsequently exploiting them upon discovery. We also combine the bacterium model in a bacterium–flock algorithm for the purposes of preventing collisions between robots or organisms in addition to providing coverage to a pollutant. By adjusting the velocity of individuals, we show that we are able to control the coverage provided by the population as a whole. Furthermore, we compare the bacterium–flock algorithm with a novel gradient-ascent-flocking algorithm and the well established Voronoi partition algorithm. Results show that bacterium–flock algorithm and the Voronoi partition algorithm are capable of adapting the distribution of the individuals of a population based upon the underlying pollutant profile while the gradient-ascent-flocking algorithm is not. This shows that the bacterium–flock and the Voronoi partition algorithms can potentially be used to track a spatiotemporal function. On the other hand, the gradient-ascent-flock algorithm has a faster convergence time in some cases with the Voronoi partition algorithm having the slowest convergence time overall.
  • Thumbnail Image
    ItemOpen Access
    Tracking and sensor coverage of spatio-temporal quantities using a swarm of artificial foraging agents
    (Elsevier, 2016-11-09) Oyekan, John; Gu, Dongbing; Hu, Huosheng
    Using a network of mobile sensors to track and map a dynamic spatio-temporal process in the environment is one of the current challenges in multi-agent systems. In this work, a distributed probabilistic multi-agent algorithm inspired by the bacterium foraging behavior is presented. The novelty of the algorithm lies in being capable of tracking and mapping a spatio-temporal quantity without the need of machine learning, estimation algorithms or future planning. This is unlike most current techniques that rely heavily on machine learning to estimate the distribution as well as the profile of spatio-temporal quantities. The experimental studies carried out in this work show that the algorithm works well by following the concentration gradient of a dynamic plume created under diffusive conditions. Furthermore, the algorithm is inherently capable of finding the source of a diffusive spatio-temporal quantity as well as performing environmental exploration. It is computationally tractable for simple agents, shown to adapt to its environment and can deal successfully with noise in sensor readings as well as in robot dynamics.