March 6, 2018
The future of unmanned aerial vehicles (UAVs) is autonomy. Giving UAVs the ability to operate on their own opens up a world of possibilities including package delivery, photography, surveillance, and more. Today, most UAVs still need someone to control it from the ground. Also, most access their computing power from a computer far below. These invisible tethers inhibit the capabilities of these “sky-is-the-limit” technologies.
A team of faculty from across Syracuse University’s College of Engineering and Computer Science are collaborating to give UAVs the ability to operate on their own—without a pilot or external source of computing.
“The moment a UAV travels beyond visual line of sight, it has to have the autonomy to navigate safely and reach its destination,” explained Associate Professor Amit Sanyal.
Sponsored by a $400,000 National Science Foundation grant, their project, “Enabling Multimodal Sensing, Real-Time Onboard Detection and Adaptive Control for Fully Autonomous Unmanned Aerial Systems,” combines the expertise of five Engineering & Computer Science professors.
Computer engineers Qinru Qiu and Jian Tang handle deep reinforcement learning and integration; aerospace engineer Sanyal builds the UAVs and oversees their controls; electrical engineer Senem Velipasalar contributes image recognition and detection algorithms; and computer engineer Yanzhi Wang constructs the necessary computing hardware.
Current UAVs are mostly remote controlled. Some have the ability to capture video for object recognition, but very few have the ability to process that information onboard for the purposes of sensing and avoiding obstacles in their path. Instead this info needs to be sent back to a headquarters to be processed, then back to the UAV. The few current models that can process info onboard are very limited in their capabilities.
The team in Engineering & Computer Science is aiming to make UAVs completely “onboard autonomous”—with computing power far greater than anything that currently exists.
“Basically, our goal is to bring artificial intelligence to UAVs,” said Qiu. “It requires a loop of sensing and processing and control. Current technology isn’t able to close that loop quickly enough during flight. Our goal is to close the loop as quickly as possible.”
Every component of their system must be customized to serve this purpose, and their work spans many technical disciplines.
“The trick is to amplify UAVs’ processing speed. Significant computation and storage are needed to accomplish this because UAVs need to perform a great deal of image processing and set and follow trajectories. This also requires a lot of power and that depletes the battery power quickly,” explains Wang. “We are building our own hardware to address these obstacles.”
It isn’t just the computing software and hardware that need to be perfected to accomplish this. The UAV itself must be suitably designed to accommodate these capabilities. That work is taking place in Sanyal’s UAV lab at the Syracuse Center of Excellence. In this facility, researchers are able to fly and test different UAV capabilities.
“This work has many layers, and each layer needs to be optimized, including how the UAVs actually flies,” said Qiu.
When all of these different pieces are combined, UAV technology will take a significant step forward toward making things like package delivery by UAV a commonplace reality.
Syracuse University is an ideal place for this work to bloom. While the College of Engineering and Computer Science studies and develops the technical capabilities of UAV, other SU schools and colleges can explore the implementation and application of this new technology in policy, business, law, public communication, and more. Plus, the Central New York region is positioned to lead the development of the unmanned aircraft industry through the Northeast UAS Airspace Integration Research Alliance (NUAIR)— a coalition of more than 200 private and public entities and academic institutions working together to operate and oversee Unmanned Aircraft System (UAS) testing, headquartered in Syracuse.
Collaborations within the College of Engineering and Computer Science, Syracuse University, and the broader community will contribute to the integration of UAS into our world. It will be the innovative spirit and technical skills of engineers and computer scientists that make it all possible.