Ryan Verbrugge

Site Autonomy Engineer · Robotics Engineer

Download PDF

Experience

Site Autonomy Engineer

May Mobility · Eden Prairie, MN

  • Autonomous vehicle maintenance and upkeep
  • Sensor calibrations for vehicle fleet
  • Handled mapping updates for Eden Prairie production fleet
  • Extensive testing for upcoming software and hardware releases

Research Assistant

Michigan Technological University · Houghton, MI

  • Conducted research spanning legged robotic control to perception and path planning systems in autonomous vehicles
  • Assisted graduate student peers on additional research topics

Research Areas

  • 8/2024 – Present: Winter Snow dataset for LiDAR systems and neural network training for vehicle detection in heavy snow environments
  • 6/2023 – Present: Automated bat counting system for White Nose Disease population study with the DNR (see the Fat Bat project)
  • 6/2023 – 9/2024: ARPA-E NextCar II, road surface analysis and data collection
  • 1/2023 – 5/2023 (main researcher): Bipedal locomotion and gait correction on low-mu surfaces
  • 8/2022 – 8/2023 (main researcher): Calculating fractional order calculus through the use of symmetric neural networks

Software Developer

Steelhead Technologies · Calumet, MI

  • Implemented improved UI design for web-based applications
  • Improved admin/user communication and interaction through back-end SQL data

Undergraduate Class Grader – Neuromorphics

Michigan Technological University · Houghton, MI

  • Rewrote labs to provide proper content information and formatting
  • Assisted students in asynchronous learning labs
  • Graded students on assigned tasks

Autonomous Simulations Intern

Hexagon – Manufacturing Intelligence Division · Novi, MI

  • Developed interfaces between simulation software and major autonomous vehicle software
  • Developed automotive simulations for autonomous vehicle testing and development
  • Produced documentation for customer support
  • Supported and assisted customer usage of simulation software

Undergraduate Lab Assistant – ROS

Michigan Technological University · Houghton, MI

  • Transferred labs from ROS Melodic to Noetic
  • Rewrote labs to provide better flow and ease of knowledge acquisition for students
  • Assisted students in learning and understanding beginning topics for ROS
  • Started creation of new lab curriculum for students in upcoming years

IT Operations Student – Tier 1

Michigan Technological University IT · Houghton, MI

Education

BS in Robotics Engineering

Michigan Technological University

Additional Projects

AutoDrive Challenge II

The AutoDrive II Challenge is a GM and SAE sponsored event in which universities receive a stock Chevy Bolt EUV and make it autonomous over five years (2021–2026). Scored challenges progressively get harder each year, spanning base-level object detection to non-GPS-based localization. Teams meet each June to compete at the University of Michigan's test track, M-City.

Roles: Michigan Tech AutoDrive Team Captain, Robotics Systems Enterprise Director, Enterprise Assistant Director, Outreach Coordinator, Lab Manager, Team Lead

Personal Contributions

  • Computer vision through use of a neural network and a camera
  • Object detection and tracking through a LiDAR sensor
  • Autonomous vehicle simulation for subsystem testing
  • Implementation of feature-level sensor fusion
  • Creation of vehicle behavior management system
  • Creation of mapping and path planning system using a standard planning algorithm
  • Built LiDAR-based localization system from scratch

Major Contributions

LiDAR Object Detection In my first year on the team, and in my first year working with this system, I worked on basic Euclidean Clustering and plane-ground filtering. Both of these implementations were done through the use of PCL.
Vehicle Navigation, Path Planning, and Behavioral Subsystems In Year 3 of the AutoDrive II Challenge, I implemented a D* Lite path planner for basic vehicle navigation through GPS-defined map infrastructure.
LiDAR-Based Localization System (See github.com/rcverbru/divining-rod) — In Year 4, teams had to navigate an environment with intermittent GPS signal drops. I was assigned to build a new localization system from scratch, adapting KISS-ICP, a simple ICP-based localization method, into a lightweight system that could reliably determine our position and navigate safely to the end goal.

Skills

Languages

C++ C Python MATLAB

Software

ROS Linux PyTorch Virtual Test Drive CARLA Unreal MATLAB DSD & RoadRunner Simulink Inventor NX

Topics of Interest

Perception Mapping & Planning SLAM Simulation Autonomous Vehicles LiDAR Camera Vision Artificial Intelligence