| Popis: |
Multirobot adaptive navigation maneuvers a multi-vehicle system based on characteristics of the environment to autonomously localize features of interest. This navigation method can be more time and energy efficient than conventional navigation methods. Most work in this area explores scalar fields, where a single characteristic value is associated with every point in the environment. This work is an initial testbed exploration of adaptive navigation for vector fields, where every point in the environment is associated with a multi-parameter value. A vector field can represent a single physical quantity such as water/air flow or multiple simultaneous and collocated scalar quantities such as temperature and gas concentration. The contribution of this work is the extension of an existing adaptive navigation testbed to support vector field representations, navigation, and further research. Vector fields are generated using a large-format printer to print 8-bit colored floor mats. Mobile robots, equipped with RGB sensors, sense the color and, through calibration, estimate the underlying vector field. This paper will walk through our process of generating vector fields and a calibration method to be used for adaptive navigation. Successful results from two adaptive navigation experiments are shown in the paper: finding a source with a single robot and using a two-robot formation to straddle a crest of high velocity flow in a vector field. |
