Research

Dr. McFall's primary research interest is artificial intelligence, having previously directed projects involving an intelligent road condition sensor and using artificial neural networks to solve boundary value problems.
 
His current research agenda involves self-driving vehicles, with an emphasis on analysis of vehicle sensing data. As an example, the video below shows the travel lane segmented with estimations for road curvature and lateral distance from lane center as well as detection of vehicles travelling is nearby lanes.
 
      
 
 
This work with videos analyzed offline has also translated into controlling real automobiles on the road. A standard KIA Optima was modified to be drive-by-wire capable where a motor controls steering with a timing belt connected to the steering column, servo motors actuate the accelerator pedal, and a cable attached to the brake pedal winds around a motor shaft (see images below). The video below shows the KIA being controlled remotely. Additionally, an all electric Wheego LiFe has been retrofitted for drive-by-wire capability with automotive grade equipment (see above).
 
Steering    Accelerator    Brake
 
 
Once drive-by-wire capable, autonomous features were added to the Optima including parallel parking assist and complete self-driving using the lane detection algorithm for steering control and autonomous braking (see below).
 
 
 
Numerous students have contributed to the autonomous vehicle labaoratory in various capacities as interns, taking research courses for credit, or as senior design projects. Please feel free to contact me if you are interested in joining the research team. I would like to thank all those who have made this possible: Michael Adeyosoye, Matteo Alessandro, Marcus Alves, Taylor Arnold, Emily Barbour, Justin Borsh, Jeffrey Briggs, Mikko Cain, Jonathan Burden, Rory Charest, Persis Charles, Andrew Combs, Josh Crane, Sparsha Eddu, Chibuzor Eduzor, Sam Epeagba, Tim Ervin, Thomas Fagan, Andrew Faulk, Tim Fisher, Kyle Fugatt, Drew Geiman, Alec Graves, Kelsey Hattam, Brice Hilkin, Duncan Hord, Will Howland, David Hudlow, Mahbubul Islam, Juan Janse Van Rensburg, Paulo Kleyzer, Matham Latif Al-Saaty, Matthew Lawrence, Steffen Lim, Nicholas Mason, Nikhil Ollukaren, Felipe Magno, Anika Marks, Rachelle McCord, Victor McKoon, Shrey Nagnur, Vivian Nguyen, Joel Perez, Tevin Phillip, Jarred Prince, Adam Ramsey, Danica Roberts, Austin Sadler, Chris Salmons, Hector Sanchez, Edward Sheeran, Yusef Skinner, Kyle Smith, Allen Stewart, Matthew Strauss, Tony Thompson, David Tran, Bryan Vincent, Trey Walston, Reid Wells, and Brad Williams.
 
Below is a list of student projects conducted in the autonomous vehicle laboratory:
  • Summer 2014
    • Develop a first-generation "go-kart" for autonomous vehicle testing
  • Fall 2014
    • Evaluate lane boundary detection software
  • Spring 2015
    • Mechanical design of second-generation go-kart for autonomous vehicle testing
    • Implement lane boundary detection software on Raspberry Pi single-board computer
    • Equip KIA Optima with actuators to control steering, braking, and acceleration
  • Summer 2015
    • Compare image processing speed with Odroid XU3 and Raspberry Pi 2 single board comptuers
    • Test fully autonomous capabilities of go-kart platform
    • Investigate algorithms for self-parking
    • Develop communication protocols for acquiring LIDAR data from Hokuyo UBG-05LN
  • Fall 2015
    • Control driving of KIA Optima by remote control with automatic braking when obstacles are detected
  • Spring 2016
    • Implement parallel parking assist on the KIA Optima
    • Create control algoirthms for maintaing desired speed and turning radius in KIA Optima
    • Develop obstacle detection and tracking algorithms using the Hokuyo UBG-05LN LIDAR
    • Assess reliability of lane detection software in real driving conditions
    • Write software to identify position on a digital map using current GPS location for path planning and fore-warning when approahcing intersections
  • Summer 2016
    • Road sign detection and distance measurement with stereo camera
  • Fall 2016
    • Build robot and begin collecting data to use convolution neural networks for training lane-keeping steering
    • Test a fully autonomous KIA Optima on a short section of road
  • Spring 2017
    • Train convolution neural networks to perform visual odometry
    • Retrofit electric car to be drive-by-wire capable for remote control from a call center
  • Summer 2017
    • Pedestrian detection and distance estimation
  • Fall 2017
    • Improve computer-controlled braking response and strength in electic car and implement automatic braking with obstacle detection
  • Spring 2018
    • Implement localization and mapping functionality using ROS nodes on mobile robot
    • Literature review and data preparation for detecting road boundaries with LiDAR data and convolutional neural networks
  • Fall 2018
    • Build small mobile robot based on NVIDIA Jetson TX2 and explore possiblities of control motion using an end-to-end solution with convolutional neural networks
    • Perform literature review of technologies used in connected vehicles
  • Spring 2019
    • Develop code to fine-tune pre-trained models to perform image classification and object detection
    • Explore strategies to perform superresolution of images in both time and resolution
    • Explore reinforcement learning for path planning of a mobile robot
  • Summer 2019
    • Development of autonomous vehicle components in CARLA simulation environment including path planning using reinforcement learning
 Publications in peer-reviewed journals:

Publications in books:

  • K. McFall, "Using Visual Lane Detection to Control Steering in a Self-driving Vehicle", Smart City 360°, Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, Vol. 166, pg. 861-873, June 29, 2016, http://doi.org/10.1007/978-3-319-33681-7_77.
  • K. McFall, An Artificial Neural Network Method for Solving Boundary Value Problems, VDM Verlag Dr. Müller, Saarbrücken 2008, ISBN 978-3-8364-5955-6

Publications in peer-reviewed conference prodeedings:

Other scholarly presentations:

  • K. McFall, keynote speaker, “Artificial Intelligence and Autonomous Vehicles” International Congress of Innovation and Technology, Barranquilla, Colombia, November 6, 2015.
  • N. Yadav, M. Kumar, K. McFall, "An Artificial Neural Network Approach for the Mass Balance of a Reactor in Steady State," 2013 SIAM Conference on Applications of Dynamical Systems, Snowbird, Utah, May 22, 2013
  • K. McFall, "Road Condition: Acoustic and Image Analysis", Transportforum sponsored by the Swedish National Road and Transport Research Institute, Linköping, Sweden, January 9, 2002
  • K. McFall, “Machine Recognition of Road Condition Using Optical Neural Informatics,” AURORA workshop, Transportation Research Board Annual Meeting, Washington DC, January 11, 2001.
  • K. McFall, "Traffic Monitoring by New Sensors", VIKING Euroregional Workshop on Monitoring, Helsinki, Finland, September 1, 1999.
  • "Machine Analysis of Road Condition Using Optical Neural Informatics", 8th World Congress on Transportation Research, Antwerp, Belgium, July 12-17, 1998
  • “Video-based Classification of Road Condition,” 9th Standing International Road Weather Congress, Luleå Sweden, March 1998

Funded external grants/donations as PI or co-PI:

  • DENSO North America Foundation, "Autonomous Vehicle Sensor Undergraduate Training", PI, $30,000, July 2019.
  • Mohawk Industries, "A Vision-Based Automated Inspection and Classification System for Carpets using Deep Neural Networks and Machine Learning", co-PI, $9152.80, Fall 2018
  • Wheego Technologies, "Donation to Mechatronics Department", PI, $15000, Fall 2016
  • Marietta Square Branding Project, "Student Scholarship", PI, $400, Fall 2015.
  • BEST Robotics Inc., Hub Development Mini-grant Program, PI, $3000, Fall 2014
  • Environmental Protection Agency, Phase I P3 Awards “Achieving Increased Photovoltaic Panel Energy Collection with Cell-Strings That Track the Sun”, co-PI, $15,000, 2014
  • Flow-through grant from the National Science Foundation, Division of Undergraduate Education, Award number 0756992, “STEP: Toys and Mathematical Options for Retention in Engineering”, PI, $18,000, July 2008
©