Comparative Analysis of Sensors in Rigid and Deformable Modular Robots for Shape Estimation

Modular self-reconfigurable robots have the ability to change their connection topology to form large networks of distributed sensors and actuators, for example to reason about gradients in the environment or the geometry of their surroundings. Past work has focused on distributed algorithms to compute the intrinsic shape of the robot. Here, we instead focus on distributed algorithms that permit planar modular robots to compute the shape of external objects in 2D. Grounded in characterization of physical sensors, we simulate shape estimation by modular robots for increasing object size and complexity. Specifically, we compare the use of connectivity and IR sensors which are commonly available on modules, and demonstrate what may be gained from deformable modules with integrated strain sensors. The algorithms we present are dependent only on motion constraints and may generalize to a wide set of modular platforms.