Towards Artificial ATRON Animals: Scalable Anatomy for Self-Reconfigurable Robots

Self-reconfigurable robots consist of simple robotic modules that play a similar role in robots as cells play in biological organisms. The vision of self-reconfigurable robots is to make an evolutionary jump for robotics analog to biological organisms that evolved from single-celled to multicelled. Such robots can autonomously self-reconfigure its modules to adjust both its behavior and morphology, to meet the requirements in the current situation. Physical, state of the art, 3D self-reconfigurable robots, e.g. [2], [4], [6], consist of up-to a few dozen, centimeter scale, modules. However, our ultimate goal is myriad-module robots that consist of thousands to trillions of millimeter or micrometer scale modules. This paper investigates the challenge of achieving functional and fast responding myriad-module self-reconfigurable robots. We propose the use of scalable anatomies inspired by biological animals to achieve such robots. As an example we introduce an anatomy for ATRON self-reconfigurable robots, which consist of anatomical parts such as muscles, bones and joints, assembled from modules in a scalable way. Fast functional response to the environment is a general requirement in most robot applications, such as locomotion and manipulation. Functional myriad-module selfreconfigurable robots may use self-reconfiguration to selfassemble, self-repair and shape-adapt, but for fast functionality, self-reconfiguration is too slow. To solve this problem the functional roles of modules may be differentiated in the robot. Figure 1(a) illustrates a robot, which consist of few modules and is able to locomote. In such a robot the modules play relatively ’broad’ roles such as feet, leg or body. But this does not scale to myriad-module robots, since the required roles must be very different, see figure 1(b). Also, from the point of applications, an increase in the number of modules should somehow benefit the functional performance of the robot. This is the great challenge of scalable functionality : How do we scale up the diversity of functionality with the number of modules?