Research and Application of Internet of Things

Wang Jin-kuang, Jing Zhang, Minfeng Yao, Shixing Li, Jie-lun LiJournal of Machine to Machine Communications 2015 · DOI

Abstract

This paper discusses the architectures, interfaces, and behaviors of an implemented intelligent device: a mobile robot. As a case study, the robot embodies the key features of the Internet of Intelligent Things (IoT): it is autonomous, mobile, and sensorial. By enabling a computer or smartphone to access and control the robot via a standard web browser, we present a robust approach to monitoring and controlling remote objects over the Internet.

Introduction

Historically, the Internet of Things (IoT) was often synonymous with Radio Frequency Identification (RFID) technology. While RFID is excellent for tracking inventory, it is fundamentally limited by one-directional communication and minimal storage. This research argues that for the IoT to reach its full potential, "Things" must evolve into intelligent, sensorial, and autonomous agents. The challenge lies in creating a cohesive architecture that bridges the gap between low-level hardware (sensors and motors) and high-level user interfaces while maintaining real-time responsiveness.

System Architecture and Methods

The researchers developed a mobile robot as a prototype for an "intelligent device." Their approach is built on a classic three-layer IoT model: Perception, Network, and Application.

Hardware Implementation

The Perception Layer serves as the robot's senses. It utilizes an array of sensors—including ultrasonic, infrared, temperature, humidity, and light—connected via I2C and General I/O to an ARM9 S3C2240 processor. This hardware stack allows the robot to perceive its environment with high granularity.

The physical prototype of the mobile robot featuring its sensor array and embedded controller.

Software and Network Integration

The Network Layer acts as the nervous system, where the robot functions as a Wi-Fi hotspot and hosts its own web server. This allows for bi-directional communication directly with the Internet. The Application Layer employs a Browser/Server (B/S) model, removing the need for proprietary apps.

Key Technical Innovations: * Shared Memory IPC: To handle data flow between low-level sensor drivers and the high-level web server, the authors implemented Inter-Process Communication (IPC) via shared memory. This allows the monitoring process to write sensor data to a virtual memory space that the web server's CGI (Common Gateway Interface) scripts can read instantly. * AJAX Integration: On the client side, AJAX (Asynchronous JavaScript and XML) ensures the web interface can update sensor readings and send control commands without page refreshes, providing a "live" feel to remote operations.

Local system software block diagram illustrating the relationship between drivers, IPC, and web processes.

Key Insights and Ideas

  • The Robot as a Server: By hosting the control interface directly on the robot, the system becomes decentralized and more resilient than architectures relying on middle-man cloud servers.
  • Bridging the Hardware-Web Gap: The use of CGI as a proxy between the web browser and physical hardware demonstrates a practical way to turn "dumb" hardware into a "web-native" object.
  • Autonomous vs. Manual Control: The system supports two modes: a manual mode for direct remote navigation and an "automatic cruise mode" where the robot uses its perception layer to navigate around obstacles independently.

Key Results

The authors successfully demonstrated a robot that could be controlled via any Wi-Fi-enabled device. Quantitative testing of the sensor suite showed high precision for an embedded system of its class:

  • Obstacle Detection: Accurate to within 1 cm.
  • Living Creature Detection: Infrared sensors detected humans or animals within 4 cm.
  • Movement Precision: Wheel speed measurement was accurate to 0.01 cm/s.
  • Environmental Sensing: Temperature accuracy of 0.1°C and humidity accuracy of 0.5%.

Limitations and Future Work

The primary limitation identified was a computational bottleneck. The 400MHz ARM9 processor struggled with heavy data tasks, specifically resulting in a video signal that was "not smooth enough" due to a lack of dedicated image processing capabilities. Future work will focus on standardization of sensor-to-processor interfaces to allow for more rapid, "plug-and-play" integration of diverse hardware components.

Summary updated

Abstract

This paper discusses the architectures, interfaces as well as behaviours of our model of an implemented intelligent device, a Robot. As a case study, the Robot has all the key features of the Internet of Intelligent Things (IoT): autonomous, mobile, and sensorial. By achieving the goal of having a computer or smart phone to access and control the Robot, we present an approach to control and monitor a remote object to move through the Internet.
BibTeX
@article{jinkuang2015research,
  author = {Wang Jin-kuang and Jing Zhang and Minfeng Yao and Shixing Li and Jie-lun Li},
  title = {{Research and Application of Internet of Things}},
  journal = {Journal of Machine to Machine Communications},
  year = {2015},
  doi = {10.13052/jmmc2246-137x.132},
}
Figure