Modeling mutual capabilities in heterogeneous teams for role assignment

Abstract

We have developed a novel type of submersible called a Synergistically Propelled Ichthyoid [SPI] which is propelled by the combined jet action and oscillatory motion of a fluttering fluid-conveying tail. Two dynamic models for an SPI are presented and solved. An analytically tractable model presented in a previous work which assumes a submersible at constant velocity has been included as a reference point. A more complex model in a general planar reference frame is also presented and is solved numerically. These simulations show a benefit to using a fluttering tail relative to a dimensionally identical rigid tail. Results which show that the SPI can be turned by an appropriate choice of time-variable fluid velocity are also presented. Construction details of an experimental realization are provided. Preliminary measurements taken using that platform are also provided; these measurements qualitatively confirm the simulation's conclusion that a fluttering flexible tail is capable of higher speed than a dimensionally identical rigid tail.

Robust Task Scheduling for Heterogeneous Robot Teams under Capability Uncertainty

“…If such a decomposition exists, the task decomposition problem is presolved and decoupled from the assignment and scheduling problem. Many previous works in coalition formation [18], vehicle routing [14], [19], job shop scheduling [20]- [22], and robotic soccer games [23], [24] deal with such non-complex (or compound) tasks. In [19], the tasks are decomposed into roles for medical agents and transport agents while connected with scheduling constraints.…”

“…Each MDP is solved beforehand, and jointly the probability of completing the tasks is maximized. In [23], [24], [43], an agent's capabilities are represented as a Gaussian random vector where the uncertainty is captured in the distribution. A task requirement is then specified by the minimum value needed for each capability type.…”

Teamwork with Limited Knowledge of Teammates

“…Other research includes Jones et al's (2006) research on pickup teams cooperating to accomplish a treasure hunt. Liemhetcharat and Veloso (2011) reason about selecting agents for ad hoc teams, and in robot soccer, Bowling and McCracken (2005) consider the case where the ad hoc agent is given a different playbook from its teammates. Further work into ad hoc teams using stage games and biased adaptive play was performed by Wu et al (2011).…”