This is the first installment of what will be a regularly updated series.

The Month of the Thesis

I officially registered for my Master’s Thesis in the month of February and spent the rest of the month easing into it by learning about the work already done. March will be the time to work on defining my project and begin work on it.

About my Thesis

I am working on my thesis under the guidance of Dr. Alireza Ramezani at the Silicone Synapse Lab (SS Lab) at Northeastern University.

I am yet to define my problem statement, but I will be working with SS Lab’s innovative robotic bat, Aerobat.

Bats are extremely agile and can fly long distances in comparison to their size, all while consuming very little energy. Contrast this with UAVs of today, where the more agile drones have a battery life in the order of minutes, and the only drones capable of flying for over an hour are very large and not very agile.

Moreover, traditional UAVs have other deficiencies in the way they fly. In confined spaces, larger drones struggle to fly as their thrust vortex is disrupted by turbulent airflow, while smaller drones have very small battery lives that get progressively smaller with each sensor added to the system.

And finally, there is the issue of safety. Despite their popularity and immense potential usefulness, UAVs are still largely banned in most populated places across the world because of the danger the propellers or the weight of the UAV poses in injuring a person in the event of a crash or a loss of control.

Aerobat solves both these issues. It’s light weight body and flexible wings make it safe to use in populated areas, and the method of thrust generation, through flapping of the wings, makes it very effective in confined spaces, much like a real bat.

This is by no means the first of it’s kind. Way back in the 1480s, Leonardo Da Vinci was already thinking about [flapping wing flight]. More recently, in 1998, California Institute of Technology successfully tested the Microbat, managing a 9 second long flight. There have been massive improvements since then, fueled by advancements in technology such as 3D printing and flexible electronics. Today, there are research groups and companies across the world attempting to perfect this class of Micro Aerial Vehicles called the Ornithopter. But while most of these designs focus on mimicking the flight of birds, bats are significantly more complex to model, but also provide much more agility.

As the only mammals capable of active flight, they have a shoulder on each wing capable of moving in all directions, an elbow, a wrist and five fingers that can grip onto surfaces for perching. Containing around 40 joints, bat wings present huge potential for precise unmanned flight, and Aerobat takes steps towards achieving that.

Initial Steps

With a prototype of Aerobat already constructed and a mathematical model developed, the next step is teaching Aerobat how to fly. In order to do that, the prototype will be mounted at the end of a robotic arm that moves it around in front of a fan array. This allows for collecting data on wind forces on the wings in a controlled environment.

This page will be updated as I have more updates on the progress of this project.

Beyond the Thesis

Along with my thesis, I am also taking two courses: Autonomous Field Robotics and Legged Robotics.

Autonomous Field Robotics

So far we have learnt the concepts behind image mosaicking, starting by making a panorama using manually selected points before moving on to feature detection and image stitching for larger datasets using GTSAM. This month, we will work on 3D Structure from Motion to build a 3D model of an object from a series of pictures from a single moving camera.

We are also studying landmark papers in computer vision and SLAM, including ORBSLAM, Binary Bag of Words, Lego Loam, Demon and Deep VIO.

Legged Robotics

Starting from the basics of Robot kinematics and control using limit cycles, we have now moved to Lagrangian Dynamics and will build up to designing a controller end to end for the NEU Husky quadruped robot. Using this controller, we will attempt basic maneuvers such as trotting, and as a challenge, attempt to balance on a ball.