Rhino: An Open-source Embedded Motherboard Design Enabling Complex Behavior of Intelligent Robots

Yucen Gao, Liang Gong, Yixiang Huang, Chengliang Liu, Xinyang Li

Abstract: In recent years, Robot Operating System (ROS) has become a de facto standard for many robotic systems. However, there lacks a general-purpose control hardware to perfectly support ROS applications in an embedded fashion. In this paper, we take a hardware/software co-design methodology and a loosely coupled design methodology to develop a ROSoriented motherboard dedicatedly for facilitating high-end intelligent robotic applications. First, orienting around the ROS computational graph level, the hardware/software co-design is proposed to realize a mirrored modular design paradigm. Second, an open-source ROS motherboard, namely the "Rhino", is accordingly designed with the highlight of accelerating the embedded neuromorphic computation. Third, real-time performance and feasibility of Rhino are validated at different scales. Experimentation shows that the open-source prototype motherboard is eligible for ROS-based robot development and outperforms the conventional IPC and tailor-made control board. ROS-oriented hardware/software codesign paradigm complements the ROS ecosystem with an open-source AI-enabled motherboard for developing intelligent robots.