George Brown College - Machining & Welding

Oct 2024 & Jan 2025

Description: A partnership between UofT and GBC allows students to have access to the MIE machine shop through courses such as:-

1. Basic Machining: Teaches machine shop safety along with how to use machine shop equipment such as lathe, mill, and drill press. In this course a students are to create a pneumatic compressed air engine using machine shop equipment listed above and instructions provided.

2. Introduction to Welding: This course focused on oxy-acetylene, stick welding, manual arc & gas metal arc welding. Through this course, I gained an understanding of the type of electrode to use, how to minimize slag, proper welding form, and other basic welding practices.

Turtlebot Mail Delivery Route

Sep 2024 - Dec 2024

Description: The TurtleBot needs to deliver mail on a closed-loop path to any of the several specified offices. The Galbraith Memorial Mail Robot project utilizes course content and techniques such as a PID control system, localization, and Bayesian tracking. The particular goal of the project is to guide the TurtleBot through an arbitrary route chosen by the teaching assistants and deliver mail to three offices.

Role: We broke down our solution into two main parts provided below.

1. Route Execution - Calibrating the RGB values of the colours was required to detect the office identity. While we initially used a PID controller for it's accuracy and speed we had to switch to a P-Controller for the errors that came up during the testing phase. The index of the light intensity which corresponded to the white line was required for following the line.

2. Localization Designing a control system to enable the Waffle Pi to deliver mail to arbitrarily chosen offices on a closed loop based on Bayesian localization required breaking the solution into multiple parts. The first part was to create a localization framework code simulating a generic Bayesian localization algorithm. This was done through state prediction and state updating. State prediction (priori estimate) was taken care of through probability distribution (previous posteriori estimate) and the control input. State update (posteriori estimate) task was based on the priori estimate and colour codes taken through prototyping. The robot had to turn 90° to simulate delivery.

---

A high-level pseudoscope is provided below for reference:-

Initialize map, probability distribution and colour codes

while True do

	i ← i − 1

	if detecting white then

		PID Control

	else if detecting colour then

		priori←state prediction #control input, posteriori estimate

		posteriori←state update #colour masurement, priori estimate

			if max(posteriori) == office location then deliver mail

			else

		continue

			end if

	end if

end while

Prospective Aerobody Designs - Gen 12

Apr 2023 - Dec 2023

Description: As the aerodynamics lead for Gen 12, I was responsible for the aerodynamic development of our solar car, designing and evaluating 50+ prospective aerobody designs to maximize performance. Focused on minimizing drag and maximizing efficiency through my optimized design process and iterative testing aiming to have a lower average simulated CdA than Gen 11 (≈0.0667). The optimized design process involved integrating aerodynamic considerations with structural, solar, and mechanical constraints to develop a balanced and high-performance design.

I also led the optimization of our approved designs, implementing more efficient simulation workflows, and automated draftability analysis. These improvements significantly reduced iteration times, allowing for faster design exploration and validation. Collaborating closely with other sub-teams, I ensured seamless integration between aerodynamics and overall vehicle design, contributing to a more streamlined and competitive solar car.