Emergent Swarm Morphology Control of Wireless Networked Mobile Robots
We describe a new class of decentralised control algorithms that link local wireless connectivity to low-level robot motion control in order to maintain both swarm aggregation and connectivity, which we term “coherence”, in unbounded space. We investigate
- PDF / 2,163,796 Bytes
- 33 Pages / 439.37 x 666.142 pts Page_size
- 1 Downloads / 193 Views
Emergent Swarm Morphology Control of Wireless Networked Mobile Robots Alan F. T. Winfield and Julien Nembrini
Abstract We describe a new class of decentralised control algorithms that link local wireless connectivity to low-level robot motion control in order to maintain both swarm aggregation and connectivity, which we term “coherence”, in unbounded space. We investigate the potential of first-order and second-order connectivity information to maintain swarm coherence. For the second-order algorithm we show that a single β parameter—the number of shared neighbours that each robot tries to maintain—acts as an “adhesion” parameter. Control of β alone affects the global area coverage of the swarm. We then add a simple beacon sensor to each robot and show that, by creating a β differential between illuminated and occluded robots, the swarm displays emergent global taxis towards the beacon; it also displays interesting global obstacle avoidance properties. The chapter then extends the idea of β heterogeneity within the swarm to demonstrate variants of the algorithm that exhibit emergent concentric or linear segregation of subgroups within the swarm, or—in the presence of an external beacon—the formation of horizontal or vertical axial configurations. This emergent swarm morphology control is remarkable because apparently simple variations generate very different global properties. These emergent properties are interesting both because they appear to have parallels in biology, and because they could have value to a wide range of future applications in swarm robotics.
10.1 Introduction This chapter investigates robot swarms that combine sensing, locomotion and morphological adaptivity. We develop wirelessly connected robots in which the wireless network becomes the “glue” physically connecting the robots. Previous work A. F. T. Winfield (B) Bristol Robotics Laboratory (BRL), University of the West of England, Bristol, UK e-mail: [email protected] J. Nembrini Media and Design Laboratory, EPFL, 1015 Lausanne, Switzerland e-mail: [email protected] R. Doursat et al. (eds.), Morphogenetic Engineering, Understanding Complex Systems, DOI: 10.1007/978-3-642-33902-8_10, © Springer-Verlag Berlin Heidelberg 2012
239
240
A. F. T. Winfield and J. Nembrini
studied the possibility of gathering sensory data across an ad hoc wireless connected network of mobile robots with range-limited communication and was focused on randomly moving robots in a bounded space [34]. Here we extend this idea to an unbounded space, controlling the behaviour of the robots in order to form a dynamically connected stable swarm, which we call a “coherent” swarm. Working primarily in simulation, but with partial confirmation of results with real-robot experiments, we show that with only an omni-directional wireless and collision avoidance device we can achieve coherence in an unbounded environment, i.e., the swarm forms a single connected communication network. Adding a simple beacon sensor, we also demonstrate emergent directed swarming (tax
Data Loading...
