ReBots: A Drag-and-drop High-Performance Simulator for Modular and Self-Reconfigurable Robots

A key challenge in self-reconfigurable robotics is the development and validation of complex distributed behaviors and control algorithms, particularly for large populations of modules. Physics-based, 3D simulators play a vital role in helping researchers overcome this challenge by allowing them to approximate the physical interactions of connected, autonomous robotic systems with one another and with their surrounding environments in a fast, safe, and low-cost manner that can reveal physical details that are critical to successful control. Current state-of-the-art self-reconfigurable robot simulators require users to have extensive programming (and software engineering) knowledge. Additionally, tasks such as translating specifications of real-world modules into simulated ones, creating complex configurations of modules, and designing complex environments are text-based, time-consuming, and error-prone tasks in these simulators, limiting their usefulness to quickly approximate real-world scenarios. This paper proposes ReBots, a drag-and-drop, high-performance self-reconfigurable robot simulator built on top of the Unreal Engine 4 (UE4) game engine. The mouse-and-keyboard GUI interface of ReBots allows users to rapidly prototype new modules, drag instances of them into environments, move and rotate modules, connect modules to one another, modify module properties, rotate module motors, change module behaviors, create complex and realistic environments, and run/pause/stop simulations. The results show that ReBots demonstrates high-performance and scalability of self-reconfigurable and modular robots with complex, distributed and autonomous behaviors in simulated realistic environments, including simulations of environments with up to 2000 autonomous modules physically interacting with one another.