Typical acoustic echo cancellation systems use linear LMS adaptive filters and cannot compensate for effects resulting from nonlinear components. I designed and implemented cascaded filters employing adaptive B-Spline and Piecewise Linear prefix blocks as means to adjust for nonlinear distortion. I proposed and demonstrated the effectiveness of a novel off-line adaptation/fitting method (based on the least squares method for linear FIR filters) for computing optimal parameter configurations in cascade architectures.
Brushless DC (BLDC) motors are popular in small autonomous vehicles because of their lower energy consumption. The standard protocol for BLDC speed control uses pulse width modulation to drive a three-phase inverter circuit, but it is also possible to control speed through rail voltage modulation. I created a controller supplementary to the three-phase inverter which maintains motor speed at varying rail voltages. Using this controller, I characterized motor energy consumption and motor dynamics as a function of rail voltage. This project supports the current development of small aerial autonomous vehicles for reconnaissance and surveillance.
The talk box is an effect pedal for the electric guitar (amongst other instruments) used notably by Peter Frampton. Via a vinyl tube, the device channels instrument sound into a user's vocal tract, upon which he can shape the output to have characteristics of intelligible speech. I used cepstral liftering to extract vocal tract response from a segment of speech, thus mimicking the pedal's capability of directly relying on the function of the user's vocal tract. Through the cepstral lifterning approach, I developed and implemented a real-time processing algorithm to digitally produce intelligibility in arbitrary carrier signals.
Affectionately known as Car Lab, this is the capstone design course of my undergraduate curriculum. The goal of this course is to create an autonomous vehicle for tasks such cruise control and navigation. There is little explicit instruction, and every team has a unique approach to the problem. Throughout the course, my lab partner, Chris Payne, and I familiarized ourselves with the necessary hardware and software components en route to a comprehensive system implementation. For our final project, unofficially titled "Simulation of a Grad Student", we created a vehicle which runs away from and stays within the bounds of a light ring while adjusting its course to avoid obstacles ("advisers") inside the ring.