The effect of morphology of a legged robot on traversing rough terrain

Abstract

When a disaster strikes a human-engineered environment, it can be difficult and dangerous for rescue workers to search for survivors. Collapsed buildings, fire and other hazards can make specific areas too dangerous to send in a human or dog. In this case, urban search and rescue (USAR) robots can be useful. Various researches have been conducted on small legged robots navigating over rough terrain. The design of these robots varies amongst researches, ranging from short four-legged robots to long flexible centipede-like models and even modular robots. Often, a new body morphology feature is tested by incorporating it in a new design and testing the performance of the robot on a rough terrain landscape. However, in this way the new feature is tested but not compared against other body morphologies on the same landscape. Therefore it is difficult to say which design is the most promising. In this thesis project, different body morphologies are tested and compared in simulation on various obstacles. A model of a modular centipede robot was designed, consisting of small bipedal units that can connect together and form a centipede of various lengths. Extensive simulations were conducted to be able to compare different connection types between the units (one or two active/passive degrees-of-freedom), and different robot lengths (changing the number of connected units). The results show that on a fractal landscape, adding degrees-of-freedom to the mechanical design of a modular centipede robot will not result in a better performance. Also, increasing the length shows only little improvement if the robots consists of more than 5 units. For a step or gap obstacle, an actively controlled pitch motion must be incorporated in the spine, preferably with a high torque. Increasing the length of a centipede robot with an actively controlled spine will enable it to cross a larger gap or climb a higher step. This thesis project has laid the ground work for the design of a new urban search and rescue robot using modular bipedal units. The results may be used as a guide for the mechanical design of such a robot.