Hello World!

 

I am Daniel and I am interested in engineering, technology and machine learning.


PUBLICATIONS

Jenniefer S. Hudson and Daniel Kolosa

"Versatile On-Orbit Servicing Mission Design in Geosynchronous Earth Orbit" , Jounal of Spacecraft and Rockets, Vol. 57, No. 4 (2020)

Analyzed the performace of a micro RF ion thruster for several CubeSat orbital maneuver strategies.

Jenniefer Hudson, Sara Spangelo, Andrew Hine, Daniel Kolosa, and Kristina Lemmer. 

"Mission Analysis for CubeSats with Micropropulsion", Jounal of Spacecraft and Rockets, Vol. 53, No. 5 (2016), pp. 836-846.

Analyzed the performace of a micro RF ion thruster for several CubeSat orbital maneuver strategies.


CONFERENCES

Scitech 2018
 

Daniel Kolosa, Dr. Jennifer Hudson

"A TFC Approach to Low-Thrust Trajectory Optimization in STK"

See "Graduate Research Assistant" for more under "Work Experience" below.


 

Scitech 2015
 

Daniel Kolosa, Dr. Jennifer Hudson

"A Linear Model for Low-Thrust Spiral Orbits and Optimal Control"

Developed a linear quadratic optimal control model to approximate spiral low-thrust spiral orbit targeting problems.
 


50th AIAA/ASME/SAE/ASEE Joint Propulsion Conference 2014

Daniel Kolosa, Sara Spangelo, Kristina Lemmer, and Jennifer Hudson

"Mission Analysis for a Micro RF Ion Thruster for CubeSat Orbital Maneuvers"

See "Publication" above for more.
 



WORK EXPERIENCE

Sr. Autonomy/AI Research Engineer

December 2021 - Present

Lockheed Martin, Palo Alto,CA

Software Developer

March 2020 - October 2021

Zoetis, Kalamazoo,MI

  • Developed a tool to load and process large genotype data using Python and C++ improving the processing speed from 7 to 3 hours.
  • Using C#, developed a web application allowing customer service to easily manage errors reported by customers.
  • Wrote Python scripts using Pandas to parse genetic datasets, and convert to a jason configuration format reducing creation time from 2 hours to 30 minutes.
  • Using .Net core, created an application to process genetic data to multiple databases reducing latency of existing systems.

Part-time Professor

August 2019 - December 2019

Department of Mechanical and Aerospace Engineering, Western Michigan University,
Kalamazoo,MI

  • Instructed students from various academic backgrounds on the fundamental engineering concepts.
  • Published weekly assignments and programming projects for students to complete in a timely manner.
  • Assisted students in understanding material covered in lecture and providing copious feedback to students about classroom performance.


Graduate Research Assistant

December 2015 - July 2019

Department of Mechanical and Aerospace Engineering, Western Michigan University,
Kalamazoo,MI

  • Wrote an engine plugin using C# to implement low-thrust trajectory optimization.
  • Developed a software package maximizing profit for a GEO robotic satellite servicer.
  • Developed a linear model to approximate the Newtonian equations of motion for low-thrust spiral orbits.
  • Applied the linear model to electric propulsion Cubesat designs to optimize for minimum fuel usage.

Software Developer, Intern

June 2014 - August 2014

Jet Propulsion Laboratory, Pasedena, California
 

  • Created a reference design for a 1.5 U and 3 U CubeSat resulting in a total cost of $50,000 and $80,000 per satellite, respectively.
  • Preformed Low-thrust mission analysis of Cubesat referencedesigns.
  • Developed satellite trajectory simulation software resulting in a well received journal publication by conducting trade studies of various combinations of micro propulsion systems for satellites.
  • Parsed and cleaned a dataset of electric propulsion systems allowing for the automation of mission analysis trade studiesof satellites with micro propulsion systems.

EDUCATION

December 2019

Ph.D., Mechanical Engineering

Western Michigan University, Kalamazoo, MI

Thesis:

A Reinforcement Learning Approach to Spacecraft Trajectory Optimization

 

December  2015

MS., Mechanical Engineering

Western Michigan University, Kalamzoo, MI

Thesis:

Implementing a Linear Quadratic Spacecraft
Attitude Control System 

 

December 2012

B.S., Aeronautical Engineering

Western Michigan University, Kalamazoo, MI


SKILLS

Programming


Python, SQL, C#, Matlab, C/C++, Java, Javascript, HTML, Bash, R, , Visual Basic
 

Operating Systems


Linux/Unix (Ubuntu, Arch Linux), Windows, Mac OS X

Software Experience


Tensorflow, Keras, Pytorch, Numpy, Scipy, Matplotlib, Scikit-Lean, Pandas, Jupyter Notebook, STK 11, Orekit, Micosoft Office, Django, .NET core

Version Control


GIT


PROJECTS

Carla pilot/vision - Developing a autonomous vehicle to navigate in a simulated environment.
 

Using the CARLA library and Pytorch, trained a car (agent) to navigate the environment using the Twin-Delayed Deep Deterministic Policy Gradients. The agent will be trained using data obtained from an expert navigating a car through the environment. The agent will be able to detect objects in its environment using the YOLOv4 algorithm and will be able to react based the objects detected. This is currently a WORK IN PROGRESS.
 

Satmind - A Reinforcement Learning Approach to Spacecraft Trajectory Optimization
 

Based off of Deep Deterministic Policy Gradients, developed a machine learning algorithm to control an electric propulsion satellite. he algorithm was able to learn 3 spacecraft missions: orbit raising, semimajor axis change (MEO to GEO), and inclination change maneuvers. Both the actor and critic are use a fully-connected neural network using layer normalization and RELU for the activation layers
Version 1.0 is avaliable!
 

Space Debris Classifier
 

Imported a space debris dataset from star-track.org listing all of the tracked space objects around Earth. Used a Random Forest Regressor model to predict the decay date of various objects and optimizing its hyperparameters using grid search cross validation to achieve a training a test accuracy of 98% and 68%, respectively. Also implemented a nearest neighbors classifier to predict an object’s radar cross-section resulting in a training and test accuracy of 70% and 68%, respectively.  



Foot Pressure Mapping

 
Using Piezoelectric sensors to map the pressure distribution on the foot. An arduino was used to process the input signals from the sensor and transmitted via bluetooth. An android app was developed to interface with the arduino allowing a user to view where more pressure was applied.


 

 

TFC Engine Plugin

 
Developed a custom engine plugin in C# for AGI's STK program. The engine plugin is based off of Thrust-Fourier Coefficients and integrates into Astrogator.
See conferences, "A TFC Approach to Low-Thrust Trajectory Optimization in STK" above, for the technical paper.


 

 

LQR Controller

 
Developed a linear quadratic regulator to aprroximate the trhsut vector for an electic propulsion spacecraft. Input an initial and target orbit state and target time and the application calculates a trajectory.




 

 

"Everything in the world is a formula that can be solved with enough persistence and creativity."

Daniel Kolosa




Email: contact@dkolosa.com
Github: github.com/dkolosa
Linkedin: Linkedin