Haptic detection of obstacles in granular media*

Navigation and obstacle detection within granular media (GM) is challenging because vision and acoustic localization methods are hindered by the granular material. In this work, we explore a strategy for detecting obstacles in GM by measuring the changes in reaction force experienced by intruders as they move near an obstacle. We first performed experiments with a rotating beam instrumented with a torque sensor at the base to measure the change in forces experienced by the beam as a function of object proximity. We tested this for five different configurations of the obstacle with respect to the beam rotation plane, and we systematically varied the obstacle distance from the rotation axis from 0 to 14 cm. We observed that when the object disrupts the upward flow of GM pushed by the beam a large change in force is observed. However, objects below or normal to the rotation plane did not result in a detectable force change. To understand this, we measured the quasi-2D flow fields through particle image velocimetry. We demonstrated the feasibility of obstacle detection in GM using an appendage-driven robot instrumented with force sensors on its limbs. Our results advance the understanding of obstacle localization in GM with many applications in robotics.