Simon Kalouche

PhD Student and Entrepreneur

Contact me: kalouche at cs.stanford.edu
Next
Next
Next

The Ohio State University

B.S. in Mechanical Engineering, 2014

Entrepreneurship & Innovation Minor
Magna Cum Laude & Honors Research Distinction
1st Place Award at Denman Undergraduate Research Forumn
Ohio State's Outstanding Undergraduate Research Award
Next

PhD Student (NDSEG)

Deep Learning for Robotics

Specializing in:

  • Deep Imitation/Reinforcement Learning
  • CNNs for Visual Recognition
  • Robot/Actuator design for force control
  • In-home and wearable robots
  • Haptics
  • Advised by Ken Salisbury and Silvio Savarese

    Projects:
    Wearable Cameras for Markerless Motion Capture using Deep Learning CNNs
    Tensorflow for Neural Style Transfer and Chatbots
    Image-Based Melanoma Classification using Deep Learning

    Next

    Master's Reseacher (NDSEG)

    Design and Control of Dynamic Force Controlled Legged Robots

    Goat is featured on (>1 Million views):
    IEEE Spectrum
    The Verge
    Tech Crunch
    Digital Trends
    The Next Web
    Wevolver
    Vocativ
    Discovery Channel, and more.

    Next

    Master's Reseacher (NDSEG)

    Compliant Control of Reconfigurable Legged Robots using Series-Elastic Actuators

    Projects:
    Computer Vision for Retail Inventory Tracking
    Stock Forecasting Using Neural Networks
    Compliant Legged Locomotion
    Window Cleaning Robot
    Actuator Design

    Next

    Undergraduate Reseacher at NASA

    Design of Compliant Mechanisms for Climbing Robots

    ACROBOT: Adhesive Climbing Robot developed at NASA JPL for servicing applications on the International Space Station.

    Next

    Undergraduate Reseacher

    Control Moment Gyroscope Stabilization of Off-Road Vehicles

    Awarded grant for $250k to develop a 1/3-scale highly manueverable off-road vehicle for the AFRL Special Ops Transport Challenege.

    Honors Undegraduate Research Thesis on the design and control of multi-axis gimbal CMGs for vehicle stabilization and Manuverability.

    Next

    Buckeye Bullet 3 Team

    AC Induction Motor Analysis


    AC Induction Motor analysis and carbon-fiber monocoque chassis studies. The 3rd Generation Buckeye Bullet uses two custom Venturi electric motors powered by LI-ON batteries; designed to safely reach 400 mph (world record). See the BB3 Website and some team press.
    Next

    Technical Skills

    Python

    • Multiple large projects using TensorFlow and Keras python libraries.
    • Used OpenCV to implement object recognition using SURF and FLANN algorithms for inventory management in warehouses and retail spaces
    • Implemented a supervised learning, deep neural network, to classify handwritten digits using MNIST Database
    • Implemented and compared performance of 3 machine learning alorithms in forcasting stock market trends using price-to-earnings (P/E) ratio, Price-to-book (P/B) ratio, Debt-Equity, Free cash flow, price/earnings to growth (PEG) ratio, and Google Trends data as input metrics for the ML hypothesis function.
    • Implemented numerous Artificial Intelligence based algorithms: 1) implemented a state-of-the-art recursive SAT backtracking solver; 2) built a (revised and dual) simplex algorithm for solving linear programs; 3) built an integer programming solver for combinatorial auctions; 4) Implemented probabilistic inference in graphical models; 5) Implemented Q-learning; 6) Built an implementation of a deep convolutional neural network for optical character recognition (OCR); 7) Implemented a state-of-the-art Counterfactual Regret Minimization algorithm for solving partial information, extensive form games like Kuhn Poker

    MATLAB

    • Used extensively for research, design analysis, and robot control in addition to course assignments in robotics and dynamics
    • Kinematic and dynamic analysis of different robot leg topologies for workspace and force envelope
    • Used for various sensor reading data analysis and data collection directly from an Arduino and Raspberry Pi
    • Simulink and SimMechanics used for robotic dynamics, linear and non-linear (LQR and Sliding Mode) controller development, and path planning.
    • Created models for a hexapod, dynamic quadruped, biped, and monopod. Designed SimMechanics controllers for hexapod walking gaits(wave, alternating tripod, strafe, etc.), and biped balancing using ankle and hip strategies as well as capture point methods. Designed controller for dynamic quadruped jumping and trotting using Virtual Model Control and Raibert style (parallel 1-DoF) controllers.

    Arduino

    • Dozens of mechatronics/robotics projects ranging from window washing robots to an internet of things devices to novel actuators.
    • Created GUI's in Processing to control motors and plot sensor data in real time.
    • Interfaced Arduino with Python and Matlab to collect and analyze data in real time using Python/Matlabs powerful matrix operations and linear algebra functions
    • Have used Arduino Uno, Mega, and Yun. In addition I've used the Pololu Orangutan and Baby Orangutan programmed in the Arduino IDE

    Solidworks

    • Over 6 years experience in an academic, research, and industry settings
    • Designed numerous mechanisms and robots including: ACRobot I and II, GOAT Jumping 3-RSR Leg, Single-axis Control Moment Gyroscope and bicycle mount hardware, continuously variable transmission (CVT) for robot actuators, variable stiffness actuator, twisted string actuator, upper-body exoskeleton, body-mounted robotic third-arm, biped feet with quad-FSR sensor array, automated catheter insertion mechanism, EDF Fan suction window cleaning robot, and more...
    • Generated detailed engineering drawings for machine shop fabrication from plastic, aluminum, steel, and titanium parts.
    • Used solidworks sketches to synthesize ACRobot I's 4-bar linkage as well as other mechanisms
    • Used Solidworks simulation and motion study toolboxes to conduct FEA on critical components and test motion of mechanisms using SW's dynamics engine

    ANSYS

    • Conducted FEA analysis on the Buckeye Bullet 3 carbon fiber monocoque and steel space-frame chassis
    • Generated Finite Element Models (FEM) for critical compressor and turbine components for GE's F110 Military Jet Engine. My analysis yielded the expected fatigue life of various stages of turbine blades.

    Altium Designer

    • Several months of tutorials desiging microcontrollers like the Arduino Nano

    TensorFlow

    • Used TensorFlow with Keras front-end to train a convolutional neural network for image-based melanoma classification.
    • Implemented Neural Style Transfer, Google Deep Dream, and a state-of-the-art chatbot using RNNs and LSTMs.
    • Fine-tuned a pretrained VGG-16 Convolutional Nerual Network.
    • Took Stanford's CS20SI course: TensorFlow for Deep Learning.

    Web Front End

    Coded www.simonkalouche.com

    Raspberry Pi

    • Built an internet of things device for retail and warehouse inventory management using a network of distributed camera modules with WiFi and iBeacon connectivity
    • Used a Gimbal Beacon to push advertising messages and sales deals to nearby customers with an iPhone
    • Uses 5 MP Pi Camera to take pictures then upload them to Dropbox using Dropbox's python API

    OpenCV

    • Used OpenCV in Python on Raspberry Pi (Raspian) to implement object recognition and image compression algorithms.
    • Preprocessed ISIC dataset to perform lesion segmentation on images of benign or malignant Melanoma. Used OpenCV to segment the mole from the background of the image and then format the image for input to a CNN.
    • Implemented SIFT/SURF, FLANN, and BF Matcher algorithms to detect items on retail shelves.

    Wolfram Mathematica

    • Wrote programs to solve kinematic equations for forward and inverse kinematics for spatial serial robots arms and parallel platforms
    • Solved constraint equations for 4-bar mechanism synthesis
    • Generated Lagrangian dynamic equations of motions for simple robot arms

    Awards and Press

    National Defense Science and Engineering Graduate Fellowship

    NSF

    NSF Graduate Research Fellowship Program

    • Received honorable mention for the 2014 class
    OSU

    The Ohio State University

    JPL

    NASA JPL/Caltech

    • Awarded 2014 NASA Ohio Space Grant Consortium Graduate Fellowship
    • Received Summer Undergraduate Research Fellowship in 2013 and 2014
    • Awarded JPL Center Innovation Fund to conduct research at Ohio State
    swartz

    James R. Swartz Entrepreneurial Fellow

    • Carnegie Mellon University 2016 Class of Swartz Fellows
    • The highly selective program develops each Fellow’s entrepreneurial potential and leadership skills through hands-on experiences, networking, mentoring and courses in entrepreneurship
    • Meet the Swartz Fellows
    CMU

    Carnegie Mellon University

    • Awarded 1st place in Carnegie Mellon’s Mechatronics 2015 Competition, $5k prize
    • Awarded James R. Swartz Fellowship
    • Awarded best applied paper at IEEE 2015 International Symposium on Safety, Security, and Rescue Robotics (SSRR)
    Pittsburgh Post-Gazette

    Pittsburgh Post-Gazette

    MSI

    Chicago Museum of Science and Industry

    The Next Web

    The Next Web

    CNET

    CNET

    Vocativ

    Vocativ

    Reuters

    Reuters News

    Discovery

    Discovery Channel

    I'm currently available for Consulting

    My areas of expertise are in deep learning and robot/actuator design

    Email: kalouche at cs.stanford.edu