1080p.mov

Impact

• Objective: Build a rover that autonomously navigates an obstacle field and mines regolith within a 10-minute run.
• Role: Autonomy team member (Year 1) → Autonomy team lead (Year 2); owned perception/localization integration and end-to-end autonomy reliability.
• Key systems: Localization pipeline evolution (vision fiducials → UWB), LiDAR perception for navigation, real-time telemetry dashboard.

What I Built

Year 1 — Autonomy Engineer

• Designed the ROS message architecture to route sensor topics into the autonomy stack (positioning + planning inputs).
• Prototyped localization using AprilTag fiducials with USB cameras for position + orientation estimates.
• Built a Python + PyQt telemetry GUI to monitor voltage, current, rover speed, and autonomy health during tests.

Year 2 — Autonomy Team Lead

• Led navigation perception using Hokuyo LiDAR point clouds to support obstacle detection and path planning.
• Integrated Pozyx Ultra-wideband localization (beacon + robot anchors) to produce robust field positioning and orientation.
• Owned integration/testing across autonomy nodes and drive control to improve run reliability under competition constraints.

System Architecture

• Autonomy runs on an Intel NUC, consuming camera/LiDAR/IMU and publishing planned motion.
• Drive node runs on a BeagleBone and sends motor commands over UART.
• Low-level obstacle detection runs on an Arduino, while an Android app provides teleop + autonomy status.