As his studies at Vanderbilt progressed, the technology concepts he faced became more and more complex. He continued to show academic excellence as he was able to understand and advance such technologies. He also honed his ability to expertly explain complex concepts to others.
For example, one of his research interests became resilient, cooperative control of networked multiagent systems, a topic not easily understood by the vast number of people, including engineers. When challenged to use his research topic to demonstrate his skill at explaining technology, he replied without hesitation:
“Essentially cooperative control is all about emergent behavior from connectedness. Whenever you have small computing nodes or dynamic systems that start to interact, they also begin to share information. You can discover emergent behaviors from the collective. The idea that you can have unexpected emergent behaviors from local interactions fascinates me. It's like the Connection piece in the entrepreneurial mindset where you can have these connections that can lead to insights, and they are insights that wouldn't have happened without the interacting components. This leads to greater value creation. Value creation can be what I would consider the emergent behavior in that case."
“My research is a microcosm of that. I study the dynamics of individual nodes and local interactions and how they can lead to emergent behaviors for the system. For example, through some very simple interactions, ducks are able to fly and maintain flying V formations. In the V formation the ducks are able to maintain good velocities and trajectories for hundreds of miles using simple local interactions. The emergent behavior there would be the flying V formation where each duck is using just basic collision avoidance approaches in order to maintain roughly the same distance from its neighbor. Each duck falls in line. They just have a couple of small rules that they're using to create the nice flying V formation.
“You could also talk about that for swarms as well as fish schools. The fish are able to avoid predators. They have little interaction rules and they just try to stay close to their neighbors without colliding with them. They are able to make the school appear as a large predator. It helps deter predators from attacking them. If they are attacked by staying close together, they make it more difficult for the predators to get any one of them.
“Another very interesting case is ant colonies. The way ants forage for food randomly, it can be shown that with these local interaction rules, behaving randomly is optimal. When you don't know where the food sources might be, random motion is optimal in terms of distance traveled by the sum of all the ants in the colony.
“So there are examples in nature. My research also applies to self-driving cars, such as swarms of cars on the expressway. I've done some research on the 802.11P IEEE standard and for communication of vehicles with infrastructure, and vehicle to vehicle."
Cyber Physical Systems
Dr. LeBlanc is also interested in Cyber Physical systems. “Cyber physical systems are more complex systems where the underlying physical dynamics of the system depend on the computational dynamics and communication layers of the system. There is an interconnectedness between the different components of the system that lead again to emergent behaviors. But now it's the cyber domain interacting with the underlying dynamics and physics of the physical system.
"Good examples are self-driving cars where you've got the vehicle to vehicle communications, vehicle to infrastructure communication, and the communication layers. You've got all of the ECUs (Electronic Control Units) and the computers on board. Those embedded processors are the computational layer and of course they've got all their physical dynamics that are determined by the actuators, the construction of the vehicle, and the safety concerns.
“All of those type of emergent behaviors depend on the quality of the communication interactions and the computation. This is what has led to so many issues in our vehicles about 10 years ago. We started having so many ECUs that it became very difficult for the car manufacturers to validate that everything was going to work as expected, when you have corner cases with different sensors failing. Auto manufacturers have gotten a lot better in their design and construction approaches, but we've also seen vehicles where you get very strange computer faults that are difficult to diagnose. They weren't necessarily constructed having in mind how many interactions you can have between the cyber layer, communication layer and the physical layer.”