Somit Kumar

I am a young engineer with a passion for innovation in the fields of aerospace, electronics, and robotics. I am currently a Junior at Colonia High School, where I am learning and practicing science through a four-year course called Science Research. I am currently on my third year of Varsity Golf, I am the founder of the Robotics and Engineering Lab, Vice President of DECA, and a Varsity Juggernaut (Academic Team member and substitute Captain). Outside of school, I enjoy working on personal projects that involve 3D printing, electronics, and coding. A few of my best projects are my interlocking drone swarm, a proprietary AI Assistant called AXIS, and a 6 DOF robotic arm. I document my work on my YouTube channel as well as in written form on this portfolio. One of my biggest goals in life is to never stop learning and inventing!

Interlocking Swarm Dynamics of Cuboid UAVs with Decentralized Heterogeneous Control Architecture

Abstract and Hypothesis

Can a UAV swarm that employs custom cuboidal frames to protect, magnets to align, and a mechanism to disengage take advantage of the versatility and aerodynamic benefits of swarming while mitigating the risks and inherent issues associated with drone swarms?

UAVs are effective systems for traversing environments aerially and deploying airborne robotics. However, they suffer from stability issues when they are small and versatile, or have low maneuverability when they are large. UAVs can rectify both problems when they are moderately sized and able to interlock into a larger aerial structure. When they are part of a swarm, standard drones may crash into each other, making control more difficult and leaving a small margin for error, increasing risk. For this reason, a drone swarm that uses cuboid frames around each drone to interlock them is least susceptible to environmental factors such as wind, or to control failures. A decentralized heterogeneous architecture means each drone serves a purpose for the swarm, but there is no 'leader' drone, increasing robustness for complex dynamic tasks. The drones interlock using magnets that are super-glued to the corner brackets attached to the frame. The tests performed on the drones will be: an obstacle course (piloted/not), testing formations (square, V-shaped, free, etc.), and, if the drones (following an algorithm) can decide their own formation to fit the task best, such as a single file line to fit through a hoop. To determine drone build quality, performance is tested between simulation/real. Finally, the drones will be tested in several real-world dynamic tasks such as opening and closing a valve, pressing a button, object retrieval (guided and unguided), and following [moving targets]. The effectiveness of the UAV swarm system will be determined by these factors, and the results will indicate if the experiment has been successful.

Page updated

Google Sites

Report abuse