Hello! My name is Marc Micatka. I am currently working as a Software Engineer at the Applied Physics Laboratory at the University of Washington. I recently finished my Master’s in Computer Science at Georgia Tech specializing in Robot Perception and Automation and am so glad to be done. Although I have to work around the busy schedule of a global pandemic, I also try and find time to travel around the country and world with my lovely wife. I enjoy finding efficient and practical solutions to problems in a class setting, a work setting, or on my own.
I love traveling and being outside, whether it’s in the mountains or in a city. Currently living in Seattle exploring the Cascades with gusto.
Experience
- Software Engineer - Applied Physics Lab, University of Washington
- Seattle, Washington January 2022 - Current
- Software engineer working on unverwater robotic manipulation and perception in challenging environmental conditions.
- Research and Development Engineer - Synthetik Applied Technologies
- Seattle, Washington May 2021 - December 2021
- Employed as a R&D Engineer working on IRD contracts and software/ML projects
- Developed software pipelines from ground up to address project and customer needs.
- Design and modify ML solutions to complex visual and audio recognition tasks.
- Optimize and deploy Azure-based IoT solutions on Nvidia Jetson platforms.
- Industrial Engineer II - Boeing
- Everett, Washington April 2019 - September 2019
- Employed as a contract industrial engineer at the Boeing factory in Everett, Washington working on the 777 aircraft.
- Experience with large scale manufacturing optimizations.
- Analyzed build plans for airplane fuselages to optimize job scheduling.
- Collected and analyzed job performance data through time studies and job observations to identify and improve assembly processes.
Publications
- Clustered Grasp Volumes for Improved Grasp Selection
- Author(s): Marc Micatka, Aaron Marburg
- Conference/Journal: OCEANS 2024 - Halifax
- Code: Github Repo
- Abstract: We propose a grasp selection algorithm that clusters grasps by pose and cost, creating discrete graspable volumes. Grasps are chosen to maximize target visibility and arm manipulability, enabling flexible autonomous or human-supervised control. The method is evaluated in simulation and with a Reach Robotics Bravo 7 arm, addressing challenges in unstructured and adversarial environments.
- PUGS: Perceptual Uncertainty for Grasp Selection in Underwater Environments
- Author(s): Onur Bagoren, Marc Micatka, Katherine A. Skinner, Aaron Marburg
- Conference/Journal: International Conference on Robotics and Automation (ICRA) 2024 (pre-print)
- Code: Website
- Abstract: We propose a framework that quantifies perceptual uncertainty in 3D reconstruction and incorporates it into grasp selection for autonomous underwater manipulation. By propagating uncertainty from multi-view reconstruction, our method improves robustness against partial and noisy measurements. The approach is validated with both simulated and real-world data.
- An Open-Source Pipeline for Automated Subsea Object Grasping
- Author(s): Aaron Marburg, Laura Lindzey, Marc Micatka, Tim Player, Anand Sekar
- Conference/Journal: OCEANS 2024 - Halifax
- Code: Github Repo
- Abstract: Subsea manipulation with ROVs is challenging due to limited perception, complex dynamics, and harsh environments. To address the high costs of development and testing, we present an open-source ROS1-based software stack for autonomous manipulation of loosely modeled subsea objects. Tested in simulation and on a low-cost platform, it serves as a baseline for advancing subsea autonomy.
Education
- 2019 - 2021
- Masters, Computer Science; Georgia Tech
Specialization: Robotic Perception and Automation
- 2011 - 2014
- BME, Mechanical Engineering; University of Minnesota - Twin Cities (Minneapolis, MN)
Technical Experience
- Programming Languages
-
Python: Extensive academic and professional experience using Python for computer vision applications, data analysis, and machine learning applications.
-
ROS/ROS2: Extensive professional experience using ROS1 and ROS2 for robotics development. Deploying on real robotic arms and autonomous underwater vehicles.
-
Gazebo: Extensive professional experience using Gazebo (classic, some experience with Ignition) to build simulation environments for robotics development. Focus on motion control simulatoin.
-
Java: Academic experience using Java for Android app development and developing software testing suites using JUnit.
-
Matlab: Academic experience using Matlab for image processing and robotics, path planning, trajectory calculation.
- Software
-
SolidWorks/Onshape: Experience using SolidWorks and Onshape for part design, developing mechanical drawings, and motion analysis for both professional and personal projects.
-
Excel: I know everyone says they’re good at Excel, but I’m actually pretty good at Excel. Professional experience designing and deploying macros and Visual Basic Applications to streamline workflows and for data analytics.
-
Cognex VisionPro: Professional experience designing visual inspection software for high-volume manufacturing lines.
Extracurriculars
- Crew Team - UMN Crew
- Eden Prarie, Minnesota May 2014 - September 2014
- Rowed competitively at regional and national competitions throughout my undergraduate years.
- Served as vice president and head of recruitment
- Won National Lightweight 4+ championship, 2014 NWIRA
Eagle Scout - Boy Scouts of America