T5: Towards UAV-Based Airborne Computing: Applications, Design, and Prototype

Presented by: Kejie Lu (U Puerto Rico), Yan Wan (U Texas), Shengli Fu (U North Texas), Junfei Xie (Texas A&M U)

Time: 9:00–12:30
Room: Kou

Abstract—In recent years, unmanned aerial vehicles (UAVs) have attracted significant attention from industry, federal agencies, and academia. To design and implement future UAV systems and applications, many researchers and engineers have been working on different UAV functions in various domains, such as control, communications, networking, etc. While all these UAV functions require advanced on-board computing capabilities, they are usually designed separately and there is a lack of a general framework to exploit airborne computing for all on-board UAV functions. In this tutorial, our objective is to address this timely and important issue by exploring a new and cross-disciplinary area: UAV-based airborne computing.

To this end, we will first systematically analyze existing and emerging UAV applications and then use case studies to demonstrate how airborne computing can help to facilitate advanced UAV functions and UAV applications. Based on such analysis, we will discuss and summarize important design guidelines for future generations of UAV systems with airborne computing capabilities. We will then introduce our recent design of a general UAV-based airborne computing platform and the latest version of our UAV-based airborne computing prototype. Finally, using our prototype, we will explain and demonstrate a number of advanced UAV functions, including reinforcement-learning based antenna heading control, image-processing based 3D mapping, and deep-learning based object detection. Finally, we will invite audience to participate in some hands-on exercises using our prototype, and we will discuss open issues and important future directions before concluding the tutorial.

Tutorial Objectives
In this tutorial, our main objective is to address a timely and important issue: UAV-based airborne computing.

The specific objectives include:
+ To give the audience an overview about the state-of-the-art UAV applications and UAV functions.
+ To discuss how airborne computing can help to facilitate advanced UAV functions, including control, communication, networking and computing functions.
+ To investigate how airborne computing can help to facilitate novel UAV application, for which we will also use case studies, such as advanced precision agriculture and emergency response, to demonstrate the potentials of airborne computing.
+ To discuss and summarize important design guidelines for future generations of UAV systems that integrate airborne computing capabilities.
+ To introduce our recent design of a general UAV-based airborne computing platform, which consists of a team of networked smart UAVs that integrate communication, control, computing and storage capabilities.
+ To explain our design and implementation of a prototype, including processor selection and carrier board design, directional antenna-enabled UAV-to-UAV communication, intelligent heading control for directional antennas, etc.
+ To demonstrate how our prototype can facilitate a number of advanced UAV functions, including (1) a reinforce-learning based long-range broadband communication, (2) on-board three-dimensional terrain map generation, (3) deep-learning based real-time object-detection, (4) advanced coded distributed computing for distributed machine learning.
+ To invite audience to participate in some hands-on exercises using our prototype.
+ To discuss open issues and important future directions.

Tutorial Outline
1. Introduction
1.1 The market trend for UAV-based applications
1.2 Regulation and policy
1.3 UAV-based applications
1.4 UAV functions and the needs for on-board computing
1.5 Motivation for UAV-based airborne computing

2. Comprehensive Analysis for UAV-based applications
2.1 A classification for UAV-based applications
2.2 A layered model for analysis and design
2.2.1 The mission layer
2.2.2 The task layer
2.2.3 The function layer
2.3 Computing-enabled UAV functions
2.3.1 Control
2.3.2 Communication
2.3.3 Networking
2.3.4 Computing
2.4 Computing-enabled UAV applications
2.4.1 UAV-based precision agriculture
2.4.2 UAV-based emergency response

3. Design guidelines and a general platform
3.1 Design guidelines

4. A general UAV-based airborne computing platform
4.1 The system overview
4.2 The components
4.2.1 The quadcopter unit
4.2.2 The control unit
4.2.3 The communication unit
4.2.4 The computing unit
4.3 The prototype
4.3.1 The processor and carrier board design
4.3.2 The directional antenna system design
4.4 Computing-enabled UAV functions
4.4.1 Reinforcement-learning based antenna heading control
4.4.2 Image processing based three-dimensional terrain map generation
4.4.3 Deep-learning based on-board object detection
4.4.4 Coded distributed computing

5. Demonstration and hands-on exercise
5.1 Demonstration of a networked airborne computing prototype
5.1.1 Processor and carrier board
5.1.2 Directional antenna system
5.1.3 Virtualization technologies
5.2 Demonstration of advanced UAV functions
5.2.1 Reinforcement-learning based antenna heading control
5.2.2 Image processing based three-dimensional terrain map generation
5.2.3 Deep-learning based on-board object detection
5.2.4 Coded distributed computing
5.3 Hands-on exercises

6. Summary, discussion, and feedback

Primary Audience
Students, researchers, and developers interested in the development of advanced UAV functions and novel UAV applications, with a background in aerospace, control, communication, networking, or computing.

Novelty
Over the past decade, UAV has been a very hot topic in multiple domains, such as aerospace, control, communications, and networking. While most recent designs require advanced on-board computing capabilities, they generally consider computing separately in different domains. Clearly, there is a lack of a general framework to exploit airborne computing for all on-board UAV functions. In this tutorial, we will address this timely and important issue by presenting a comprehensive tutorial for UAV-based airborne computing.

Biography
Dr. Kejie Lu is a professor in the Department of Computer Science and Engineering, University of Puerto Rico at Mayagüez (UPRM). He received his Ph.D. degree in Electrical Engineering from the University of Texas at Dallas in 2003. Since July 2005, he has been a faculty member in UPRM. His research interests include architecture and protocol design for computer and communication networks, cyber-physical system, network-based computing, and network testbed development.

Dr. Yan Wan is currently an Associate Professor in the Electrical Engineering Department at the University of Texas at Arlington. She received her Ph.D. degree in Electrical Engineering from Washington State University in 2009. From 2009 to 2016, she was an assistant professor and then an associate professor at the University of North Texas. Her research interests lie in developing fundamental theories and tools for the modeling, evaluation, and control tasks in large-scale dynamic networks and cyber-physical systems.

Dr. Shengli Fu is currently a professor and the Chair in the Department of Electrical Engineering, University of North Texas (UNT), Denton, TX. He received his Ph.D. degree in Electrical Engineering from the University of Delaware, Newark, DE, in 2005, before he joined UNT. His research interests include coding and information theory, wireless communications and sensor networks, aerial networks, and drone systems design.

Dr. Junfei Xie is an Assistant Professor at the Department of Computing Sciences of Texas A&M University – Corpus Christi. She received her Ph.D. degree in Computer Science and Engineering in 2016 from University of North Texas. Her current research interests include airborne networks, unmanned systems, spatiotemporal data mining, dynamical system modeling and control, and complex information systems.