Date of Award

May 2026

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Mechanical Engineering

First Advisor

Ryan R. Striker

Abstract

In the field of autonomous vehicle (AV) research, developers often face a significant gap between low-power hobbyist kits and cost-prohibitive industrial platforms. This thesis presents the design, integration, and validation of "Das Mueller," a high-performance, 1/10th-scale unmanned ground vehicle (UGV) designed to provide a transparent and cost-effective testbed for neural network development. By bridging the interoperability gap between the Donkey Car software framework and the 3D-printable Tarmo5 mechanical chassis, this project establishes a modular architecture centered on the NVIDIA Jetson Orin NX single-board computer.

The research utilizes a Model-Based Systems Engineering (MBSE) approach and rapid prototyping to address critical challenges in hardware integration and operational safety. Key technical contributions include the development of a custom mounting adapter to unify disparate hardware ecosystems and the engineering of a dual-rail power distribution system (3S for drivetrain, 4S for logic) to provide an untethered power supply. To ensure safe teleoperation at high speeds, a 20Hz network heartbeat failsafe was implemented, effectively mitigating the "runaway" risks inherent in wireless control links by reverting the vehicle to a neutral state in case of signal loss.

The system's performance was validated through a rigorous testing protocol covering component-level I2C/GPIO handshakes, PWM calibration, and field testing in various environments. While the initial build cost reached $1,570.30 due to high-tier donor hardware, the study identifies clear pathways to reduce the cost to approximately $712 without sacrificing core functionality. The results demonstrate that the integrated platform offers superior mechanical transparency and durability compared to commercial-off-the-shelf alternatives. By leveraging the high impact resilience and fatigue resistance of 3D-printed components, this work establishes a 'known-good' configuration for researchers and provides a robust hardware foundation for advancing autonomous vehicle algorithms.

Download

Share

COinS