[Paper Review] The NAO Backpack: An Open-hardware Add-on for Fast Software Development with the NAO Robot

Matías Mattamala, Gonzalo Olave+3 more|arXiv (Cornell University)|Jun 20, 2017

Robotics and Sensor-Based Localization5 references4 citations

TL;DR

The NAO Backpack is an open-hardware add-on that enables fast, external processing of computationally intensive robotics algorithms on the NAO robot by mounting an ODROID-XU4 board in a 3D-printable backpack, maintaining robot mobility and autonomy. It supports real-time sensor access via a ROS-compatible software bridge and enables efficient development in applications like visual SLAM and deep learning-based vision.

ABSTRACT

We present an open-source accessory for the NAO robot, which enables to test computationally demanding algorithms in an external platform while preserving robot's autonomy and mobility. The platform has the form of a backpack, which can be 3D printed and replicated, and holds an ODROID XU4 board to process algorithms externally with ROS compatibility. We provide also a software bridge between the B-Human's framework and ROS to have access to the robot's sensors close to real-time. We tested the platform in several robotics applications such as data logging, visual SLAM, and robot vision with deep learning techniques. The CAD model, hardware specifications and software are available online for the benefit of the community: https://github.com/uchile-robotics/nao-backpack

Motivation & Objective

• To address the computational limitations of the NAO robot’s onboard processor for complex robotics algorithms.
• To maintain the robot’s mobility and autonomy while offloading intensive processing to an external platform.
• To provide a low-cost, open-source solution for rapid prototyping and testing of advanced robotics software.
• To enable real-time access to the robot’s sensors through a software bridge between B-Human’s framework and ROS.
• To support diverse robotics applications such as visual SLAM, deep learning, and data logging using external computation.

Proposed method

• Designing a 3D-printable backpack form factor to securely mount an ODROID-XU4 board on the NAO robot.
• Integrating the ODROID-XU4 as an external processing unit to run computationally heavy algorithms independently of the robot’s onboard system.
• Establishing a ROS-compatible software bridge to enable low-latency access to the robot’s sensors from the external platform.
• Utilizing the B-Human framework’s sensor interface to synchronize data with the external processing unit for near real-time performance.
• Ensuring power and data connectivity between the NAO robot and the external board via custom wiring and power management.
• Providing open access to CAD files, hardware schematics, and software stack to promote community adoption and replication.

Experimental results

Research questions

• RQ1 Can an open-hardware backpack solution effectively offload computationally intensive robotics algorithms from the NAO robot while preserving its mobility?
• RQ2 To what extent does the software bridge enable real-time access to the robot’s sensor data from an external processing unit?
• RQ3 How does the platform perform in practical robotics applications such as visual SLAM and deep learning-based vision?
• RQ4 What is the impact of external processing on the robot’s autonomy and system responsiveness?
• RQ5 Can the platform be replicated and extended by researchers and developers using open-source resources?

Key findings

• The NAO Backpack successfully enables external execution of computationally demanding algorithms, such as visual SLAM and deep learning-based vision, without compromising the robot’s mobility.
• The software bridge provides near real-time access to the robot’s sensors, achieving low-latency data transfer compatible with ROS-based workflows.
• The platform supports diverse robotics applications, including data logging, visual SLAM, and deep learning inference, demonstrating broad applicability.
• The open-hardware and open-source design allows for full replication and extension by the research community, with CAD models, hardware specs, and software available online.
• The solution significantly reduces development time for advanced robotics software by decoupling algorithm execution from the robot’s onboard constraints.
• The integration of the ODROID-XU4 provides sufficient processing power to handle complex algorithms while maintaining energy efficiency and portability.