[Paper Review] Smartphone-based Home Robotics

Lojain Jibawi, Saoussen Said+2 more|arXiv (Cornell University)|Mar 6, 2018

Social Robot Interaction and HRI5 references3 citations

TL;DR

This paper proposes a smartphone-based humanoid robot architecture that uses off-the-shelf smartphones as the primary computing unit, reducing costs by 3x while enabling full compatibility with iOS and Android. By leveraging smartphone hardware and software ecosystems, the approach aims to accelerate robot app development and mainstream adoption through a familiar, scalable platform.

ABSTRACT

Humanoid robotics is a promising field because the strong human preference to interact with anthropomorphic interfaces. Despite this, humanoid robots are far from reaching main stream adoption and the features available in such robots seem to lag that of the latest smartphones. A fragmented robot ecosystem and low incentives to developers do not help to foster the creation of Robot-Apps either. In contrast, smartphones enjoy high adoption rates and a vibrant app ecosystem (4M apps published). Given this, it seems logical to apply the mobile SW and HW development model to humanoid robots. One way is to use a smartphone to power the robot. Smartphones have been embedded in toys and drones before. However, they have never been used as the main compute unit in a humanoid embodiment. Here, we introduce a novel robot architecture based on smartphones that demonstrates x3 cost reduction and that is compatible with iOS/Android.

Motivation & Objective

• To address the fragmented robot ecosystem and low developer incentives that hinder mainstream adoption of humanoid robots.
• To reduce the cost and complexity of humanoid robotics by repurposing existing smartphone hardware and software.
• To enable seamless integration of mobile app ecosystems (iOS/Android) into humanoid robots for faster app development.
• To demonstrate a viable, low-cost alternative to traditional robot architectures using consumer smartphones as the central processing unit.
• To bridge the gap between smartphone capabilities and humanoid robot functionality through a unified, scalable architecture.

Proposed method

• Integrating a standard smartphone as the main computational core of a humanoid robot, replacing dedicated robot processors.
• Using the smartphone's existing sensors (camera, IMU, GPS), communication modules (Wi-Fi, Bluetooth), and processing power for robot perception and control.
• Developing a hardware interface to connect the smartphone to robot actuators, sensors, and power systems.
• Leveraging Android and iOS software stacks to run robot applications natively, enabling access to 4M+ existing mobile apps.
• Designing a modular mechanical structure that securely mounts the smartphone while allowing for sensor and actuator integration.
• Implementing a real-time control pipeline that offloads perception and decision-making to the smartphone, ensuring responsiveness.

Experimental results

Research questions

• RQ1 Can a smartphone serve as the primary computing unit in a humanoid robot, replacing dedicated robot hardware?
• RQ2 To what extent can smartphone software and app ecosystems accelerate robot application development?
• RQ3 What is the cost reduction achievable by using a smartphone as the core processor in a humanoid robot?
• RQ4 How does the performance of a smartphone-based robot compare to traditional humanoid systems in real-world tasks?
• RQ5 What are the technical and architectural challenges in integrating smartphones into humanoid robot bodies?

Key findings

• The proposed smartphone-based architecture achieved a 3x reduction in overall system cost compared to traditional humanoid robots.
• The robot successfully leveraged existing smartphone sensors and processing capabilities for real-time perception and control tasks.
• The system demonstrated full compatibility with both iOS and Android, enabling deployment of standard mobile applications on the robot.
• The integration of smartphone hardware and software enabled rapid prototyping and deployment of robot applications.
• The approach supports a scalable, extensible platform that can evolve with advancements in smartphone technology.
• The architecture enables developers to use familiar mobile development tools and frameworks, significantly lowering the barrier to entry.