Controls Optimization and Networks Lab

As part of UT Dallas Controls Optimization and Networks Lab (CON lab), I have been working on using different simulation environments like Airsim (Aerial Robotics and Simulation) and Carla for developing and simulating control algorithms on autonomous Cars and drones in a high fidelity environment.

CARLA is an urban driving simulator developed for autonomous driving research. We are using it to simulate uncertainty propagation in risk based motion planning. We implement perception uncertainties into planning by using a non linear model predictive control for steering law, camera sensor model and an unscented Kalman filter for state estimation. Carla was used as it provided the basic framework to build our simulation on top.

Airsim is specifically being used for drones since it provides a steppable clock and both software and hardware in the loop features. I developed a cascaded LQR controller that provided low level controls for the drone which could then be used by the lab for simulating various control methods. Airsim, including this controller, is then being used for simulating a data driven distributionally robust trust based consensus model on a multi drone setup hovering together in a given formation as shown in the image.