Island University Team Works to Design UAVs that Can Fly as a Flock

By Darrell J. Pehr | Published: December 18, 2019

Island University Team Works to Design UAVs that Can Fly as a Flock
Texas A&M University-Corpus Christi Assistant Professor Dr. Luis Garcia Carrillo leads an Unmanned Aerial Systems Summer Program.

CORPUS CHRISTI, Texas – Birds fly together with a natural rhythm that makes the flock seem to move as one unit.

Quick decisions are being made constantly about who leads, how fast they go, how much space they keep between one bird and the next, when it’s time to slow down, and how they will all eventually come to a landing safely and in good order.

So how can the same synchronized flight be achieved in robotic unmanned aerial vehicles?

That’s a question that fascinates Dr. Luis Garcia Carrillo, Assistant Professor of Engineering in the School of Engineering and Computing Sciences at Texas A&M University-Corpus Christi.

“How can we create robots that can swarm in an autonomous way?” Garcia Carrillo said. “How can we design a robot to be aware of its neighbors, and to coordinate actions so that they are not going to be crashing into each other while doing a mission?”

Velocity, direction, and distances (between neighbors) can be computed using a mathematical approach, and a group of UAVs can fly together as a flock.

However, not only must airborne robots be capable of flying with others, they also must be able to adapt to changes, such as windy conditions, the appearance of other flying objects or an unexpected movement by another robot in the swarm. And the biggest challenge of all is the need for each flying robot in the swarm to make these changes individually and instantaneously.

To achieve instantaneous responses, scientists have studied how emotions are communicated in the human limbic system. For example, if a person touches a hot surface, they immediately draw their hand away. The signal that the surface is dangerous and to be avoided is a message that is delivered and learned immediately. Such speed in messaging and learning is crucial for in-flight adjustments by flocking UAVs, also known as multiagent systems.

Garcia Carrillo said a computational model already exists that mimics parts of the brain known to produce emotion. However, important details about how it works are not fully understood. Garcia Carrillo is working on a project that would develop a novel mathematical model of how the brain’s limbic system works, which will be used to develop a Brian Emotional Learning Intelligent Controller for multiple UAVs.

Garcia Carrillo’s work has attracted support from the U.S. Army in the form of an initial, one-year Short-Term Innovative Research (STIR) grant to develop preliminary results.

“Dr. Garcia Carrillo’s research offers an innovative approach to increasing autonomy in multi-agent systems by making use of cues associated with the emotional state of the human operation,” said Dr. Derya Cansever, program manager of multiagent network control at the Army Research Office, an element of the U.S. Army Combat Capabilities Development Command. “During collaborative tasks, humans tend to adjust their behavior based on emotion-induced sensory inputs from others. Incorporating this approach could enhance efficiency of autonomous and human-assisted multiagent systems, which will be an important operational capability for the Army.”

The project also attracted on-campus support from the Division of Research and Innovation.

Garcia Carrillo, post-doctoral researcher Ignacio Rubio Scola, and graduate student Gabriel Alexis Guijarro Reyes  are working on the project. Garcia Carrillo and Rubio Scola are working on mathematical proof of the controller. So far, the team has theoretical results and has submitted a paper for publication. Garcia Carrillo said one next step would be guaranteed performance – not only that a controller is stable, but that it can deliver certain results.

With the one-year STIR project coming to a close at the end of the year, the team is planning to report their findings and seek additional funding for more advanced work.