top of page

Jacob Higgins 

I am currently a Ph.D. student in Electrical and Computer Engineering at the University of Virginia. I work in Autonomous Mobile Robots Lab and Link Lab under the supervision of Prof. Nicola Bezzo. I earned my B.S. and M.S. in Physics also at the University of Virginia.

​

My research focuses on Optimal Control Theory, Model Predictive Control (MPC), motion planning in uncertain environments, real-time scheduling and aerial robotics.

Research

IMG_0673.jpg

Negotiating Uncertain and Occluding Environments

Autonomous mobile robots can provide invaluable help for day-to-day operations inside an office building or warehouse setting. As they travel, they must know how to share the space with other moving actors without crashing, whether it be humans or other robots. There has been much research on how to avoid moving obstacles that the robot can sense, but what about moving obstacles that the robot can't sense? One can think of a situation where a person appears from around a corner -- how should a robot handle situations like these?

​

This problem combines safety and visibility that has never been explored before, and is one area of research that I am currently focused on.

​

Read more about this work here.

intro1 (1).png

Offloading Intense MPC Computation for UAVs

Most UAV design must balance between adding devices necessary for autonomous flight (e.g., cameras) and reducing the weight to maximize flight time. This means that the computational power on-board the UAV is often limited, and must be shared among all processes required to complete its programmed tasks. One such safety-critical task is control, and an increasingly popular choice of controller is Model Predictive Control (MPC), which can be used when UAV is required to autonomously perform aggressive maneuvers (among other things). But this ability that a MPC confers comes at heavy computational cost, taking away the limited computational resources from other mission-critical processes.

​

One answer to this problem is offloading the MPC process to a nearby robot/base-station, either because this other robot is not currently using all of its computational resources, or because it simply has a more powerful computer. Currently, I am implementing a framework that can offload an MPC for the Asctec Pelican quadrotor in our lab.

ICCPS21.png
bottom of page