EMERGE Modular Robot: A Tool for Fast Deployment of Evolved Robots

Area of Science:

• Robotics
• Reconfigurable Hardware
• Evolutionary Computation

Background:

• Developing adaptable robotic systems is crucial for diverse applications.
• Existing reconfigurable hardware platforms often lack a full cycle process for morphology evolution.
• Bridging the simulation-to-reality gap in robotic morphology is a significant challenge.

Purpose of the Study:

• To present the EMERGE (Easy Modular Embodied Robot Generator) platform for full cycle morphology evolution in reconfigurable hardware.
• To detail the mechanical design, assembly, and testing procedures for EMERGE modules.
• To demonstrate the feasibility of autonomous, full cycle evolution of robot morphologies in physical hardware.

Main Methods:

• Designed and built EMERGE modular robot components using off-the-shelf and 3D-printed parts with magnetic connectors.
• Developed a visual guiding tool with AprilTag markers for accurate manual assembly.
• Evolved 30 morphologies in simulation, transferred them to physical hardware, and tested their performance, including autonomous assembly/disassembly.
• Investigated the role of magnetic connectors in protecting modules from damage.

Main Results:

• EMERGE modules are fast to build (30 min/module) and assemble (<5 min for <10 modules).
• Physical EMERGE morphologies accurately reproduced simulated performance, accounting for the reality gap.
• Magnetic connectors provided a safety mechanism against high external torques.
• Autonomous module attachment/detachment using a robotic manipulator was demonstrated.

Conclusions:

• The EMERGE platform successfully enables a full cycle process for evolving robot morphology in physical reconfigurable hardware.
• The system demonstrates the potential for fully autonomous, adaptive robotic systems.
• EMERGE offers a viable approach to balancing reusability and morphological variability in physical robot evolution.