Neural and Rehabilitation Engineering Labhttps://uwaterloo.ca/neural-rehabilitation-engineering-lab/- Working on autonomous wheelchairs for quadriplegic patients

Formerly Kindred.AI

- Configured an MPPT driver to utilize high-resolution pulse width modulation (HRPWM) written in C on a Texas Instruments Sitara chip, increasing resolution by 98% and reducing electrical noise within the system- Validated ePWM peripheral changes on bench with an oscilloscope checking against schematics on Altium, verified registers on TI CCS and further confirmed various signals with a Typhoon HIL scope and schematic editor- Built and flashed firmware using the UDS protocol on a CAN bus, observing changes on PCAN-Explorer 5 and debugged firmware using JTAG and GDB to set breakpoints, see the backtrace, and check registers- Developed and optimized a data buffering library, increasing storage capacity by 300% through type changes, scaling factors, and bitwise manipulation for efficient value storage in C- Engineered and coded innovative solutions to seamlessly migrate and update features from retiring chips, including feature setting algorithms and adding XCP to allow buffer logging- Created a pipeline to generate C firmware code directly from Simulink electrical schematic diagrams using Python, eliminating development time and removing human translation errors with tokenization and lexical analysis

- Developed an algorithm to map moving objects and predict their trajectories in 3D space for an autonomous mobile robot using C++, CUDA, OpenCV, NVIDIA’s Visual Programming Interface, and the Point Cloud Library (PCL), relying on only camera systems in a high-activity factory environment to ensure user safety when the robot is operating- Integrated with the ORB-SLAM3 mapping algorithm to localize the vehicle in a memorized map, prohibit mapping moving objects, and to control the vehicle to respond to predicted movements of objects within its environment- Engineered image processing techniques for computer vision using CUDA on NVIDIA Jetson Orin hardware to increase analysis speed by over 99% using parallel computing architecture- Designed and implemented interfaces to control the vehicle with React.JS, using a Django backend in Python to control the vehicle remotely over WebSocket’s with Django Channels- Worked with the Clang Abstract Syntax Tree and LLVM in C++ to analyze source code and automatically generate equivalent code in Python and Java to integrate with the autonomous robot

- Developed internal tools to automate client requests with Python and XML integration, reducing processing times by 95% and completely eliminating human error involved in the requests.