Experimental Implementation and Verification of Scalar Field Ridge, Trench, and Saddle Point Maneuvers Using Multirobot Adaptive Navigation

Abstract

Adaptive navigation of scalar fields is a compelling capability in which mobile robotic systems make real-time navigation decisions based on sensed measurements of the environment. This capability can enable efficient identification and location of specific features of interest within the field of interest, potentially saving time and energy while also being responsive to changing conditions. Applications can include finding the sources and impact zones of a pollutant, establishing hazard perimeters, finding safe zones, and safe paths of travel. This paper presents new work that experimentally verifies several adaptive navigation control policies for moving to/along critical scalar field features with a group of mobile robots. Specifically, we demonstrate the use of a five robot cluster of sensor-equipped mobile robots to descend ridges within a scalar field, to ascend trenches, and to move to and hold a position at saddle points. This is done through the use of differential measurements across the cluster's formation baselines and control laws that have been previously demonstrated in simulation. This paper also incorporates a new state machine within the adaptive navigation control architecture in order to monitor the performance of the individual control primitives and to respond to conditions such as losing track of the feature of interest. Finally, this paper is the first in which we have experimentally demonstrated control of a five robot group of robots using our cluster space control methodology. The experiments were conducted using a novel indoor multi-robot testbed with the ability to establish customizable scalar fields printed in greyscale on large sheets that are actively sensed by the robots to enable controlled experimental evaluation. Four different fields are used in this study in order to demonstrate the new capabilities of interest.

Keywords

• Adaptive navigation
• adaptive sampling
• differential control
• multi-robot formations
• formation control
• cluster space control