Simulation Study on the Locomotion Algorithm of Variable Topology Truss Robot based on Motion Primitives

Abstract

Variable Topology Truss (VTT) is a modular robotic system, which is an extended robot concept based on the Variable Geometry Truss (VGT). It is composed of modular members and nodes and can operate in a rough and cluttered environment. By changing the length of each member, the VTT robot can change size and shape. The VTT can also perform locomotion by rolling. Travel over a long distance can be accomplished by repeating the rolling motion, so a motion primitive is used. This reduces the computational work because the repetitive motion only needs to be calculated once. In this paper, an algorithm is presented to derive a motion primitive for an octahedral VTT. The algorithm consists of three phases: placing one node on the ground, moving the center of mass, and returning to the initial shape. On each time step, a set of possible robot motions is evaluated based on a cost function. The lowest cost motion is selected on each time step to form the optimized motion primitive. The resulting motion primitive is analyzed and validated.