Exportar Publicação

Duarte, M., Gomes, J., Costa, V., Oliveira, S. M. & Christensen, A. L. (2016). Hybrid control for a real swarm robotics system in an intruder detection task. In Squillero G., Burelli P. (Ed.), 19th European Conference on Applications of Evolutionary Computation, EvoApplications 2016. (pp. 213-230). Porto: Springer.

M. Duarte et al.,  "Hybrid control for a real swarm robotics system in an intruder detection task", in 19th European Conf. on Applications of Evolutionary Computation, EvoApplications 2016, Squillero G., Burelli P., Ed., Porto, Springer, 2016, vol. 9598, pp. 213-230

@inproceedings{duarte2016_1775764790246,
	author = "Duarte, M. and Gomes, J. and Costa, V. and Oliveira, S. M. and Christensen, A. L.",
	title = "Hybrid control for a real swarm robotics system in an intruder detection task",
	booktitle = "19th European Conference on Applications of Evolutionary Computation, EvoApplications 2016",
	year = "2016",
	editor = "Squillero G., Burelli P.",
	volume = "9598",
	number = "",
	series = "",
	doi = "10.1007/978-3-319-31153-1_15",
	pages = "213-230",
	publisher = "Springer",
	address = "Porto",
	organization = "",
	url = "https://link.springer.com/chapter/10.1007/978-3-319-31153-1_15"
}

TY  - CPAPER
TI  - Hybrid control for a real swarm robotics system in an intruder detection task
T2  - 19th European Conference on Applications of Evolutionary Computation, EvoApplications 2016
VL  - 9598
AU  - Duarte, M.
AU  - Gomes, J.
AU  - Costa, V.
AU  - Oliveira, S. M.
AU  - Christensen, A. L.
PY  - 2016
SP  - 213-230
SN  - 0302-9743
DO  - 10.1007/978-3-319-31153-1_15
CY  - Porto
UR  - https://link.springer.com/chapter/10.1007/978-3-319-31153-1_15
AB  - Control design is one of the prominent challenges in the field of swarm robotics. Evolutionary robotics is a promising approach to the synthesis of self-organized behaviors for robotic swarms but it has, so far, only produced been shown in relatively simple collective behaviors. In this paper, we explore the use of a hybrid control synthesis approach to produce control for a swarm of aquatic surface robots that must perform an intruder detection task. The robots have to go to a predefined area, monitor it, detect and follow intruders, and manage their energy levels by regularly recharging at a base station. The hybrid controllers used in our experiments rely on evolved behavior primitives that are combined through a manually programmed high-level behavior arbitrator. In simulation, we show how simple modifications to the behavior arbitrator can result in different swarm behaviors that use the same underlying behavior primitives, and we show that the composed behaviors are scalable with respect to the swarm size. Finally, we demonstrate the synthesized controller in a real swarm of robots, and show that the behavior successfully transfers from simulation to reality.
ER  -