In addition to my M.S. Thesis and Ph.D. work, I have completed a number of other interesting projects. A brief synopsis of these projects is given below, along with resulting publications and some fun pictures and videos. Please feel free to contact my with questions or comments regarding any of these projects.
Projects at Georgia Tech:
|Projects at Rice:|
|Haptic Surgical Teleoperation
Haptic Stroke Rehab
Carnival Track Game
SNS Remote Handling
1. Maximizing network throughput for over-actuated systems – 2011-present
(work performed with Timothee Cazenave and Professor Eric Feron)
Addressing a large number of actuators can be problematic in terms of bandwidth allocation. We developed an approach where it is possible to address all of the system’s actuators with only few messages, assuming that the actuators are connected via a network with distributed computation.
The method approximates the nominal control law based on the singular values of the input controllability Gramian of the system. Some initial simulations based on a series of ten mass-spring-dampers (shown below) demonstrates the effectiveness of the technique. Using only three inputs, near perfect tracking was obtainable.
Cazenave, T., R. Winck, E. Feron, W. Book, “Control Technique for Network Throughput Maximization of Over Actuated Systems,” AIAA Guidance, Navigation, and Control Conference, August 2012. (submitted)
2. Advanced hand controllers for excavators – 2011-present
(work performed with Mark Elton, Michael Valente and Professor Wayne Book)
Redesigned excavator hand controllers to be more intuitive, see image below for an example hand controller. Currently in the processes of performing human testing using an excavator simulator developed by Mark Elton. The goal is to improve operator efficiency by making the interface easier to use. Previous work has shown improvement using coordinated position control instead of the traditional joint rate control. However, coordinated position control presents added challenges such as operator fatigue and biodynamic feedthrough. The hand controllers developed in this project seek to maintain the added benefits of more intuitive control while mitigating the potential negative effects.
Click here for a video of the excavator simulator that is being used to test the hand controllers effectiveness.
3. Mapping operator inputs and actuator forces to actuator velocities for a hydraulic excavator – 2009
(special thanks to Incova and Husco International for excavator data)
Hydraulic excavators are some of the most commonly used construction machines. While years of research have highly optimized the machines in terms of controllability, the efficiency of excavators has been overlooked in the process. Now, research is focusing more on efficiency, and the method by which hydraulic flow and power are controlled is being changed in an effort to improve efficiency. One challenge of attempting to increase efficiency is maintaining the current machine controllability. This is made difficult by the internal dynamics of the hydraulic systems, which are highly coupled between the different degrees of freedom. For this project, a neural network was used to create a mapping from the operator commands and applied forces to the velocity of the boom. The results showed that neural networks could effectively create a mapping using only a limited set of training data.
4. Automated sewing machine prototype – 2009-2010
(work performed with Dr. Philip FitzSimons and Dr. Steve Dickerson)
After completing my M.S. thesis, I did some consulting work for, Softwear Automation Inc., the company that had sponsored my thesis work. The work was an extension of my thesis that primarily involved creating a more portable prototype automated sewing machine.
5. Machine vision thread tracking – 2007
For a Machine Vision course project I wrote a program in Matlab to detect threads in fabric. The fabric I used to test was denim. Below is one of the test images. The detected threads are labeled by red stars. False positives are labeled using a green circle.
6. Motion control course laboratory – 2009-2010
(thanks to Rockwell Automation for funding and equipment and to my predecessor Brian Post)
I got the opportunity to help my advisor, Professor Wayne Book, develop lab exercises for the Motion Control course that he created at Georgia Tech. It’s a cross-listed course for undergraduates and graduates. I created labs for DC motor control, trajectory planning, and state space control of large hydraulic manipulators. A more unique lab I created involved command shaping a linear stage to remove vibration from the table to which the stage was mounted. In addition to making the handouts and equipment, I taught the labs and oversaw student’s projects.
7. ABCycle: An anti-lock brake bicycle – 2009
(worked on the project in a team with Chen Shu Ngoo and Ken Marek)
We created a prototype active anti-brake system for a bicycle and experimentally verified improved bike handling during rear wheel lock. The system used hall effect sensors to detect when the wheels locked and used pneumatic pistons to disengage the brakes. A microcontroller on the bike took care of all the computation. The system was added to a freebie abandoned bike from the Bike Zoo in Knoxville, TN. As the test pilot for our experiments I can attest that it worked really well. However, although both the front and rear brakes had working ABS, I wasn’t brave enough to run experiments comparing the results of locking the front wheel with and without ABS.
Click here for a video of an early test run.
Winck, R., C. Ngoo, K. Marek, “Active Anti-lock Brake System for Low Powered Vehicles Using Cable-Type Brakes,” SAE World Congress, April 2010.
8. Modeling a gravity powered ping-pong ball launcher – 2008
(worked on the project in a team with Aaron Scott, Ben Beck, and Jay Johnson)
We used Dymola and Modelica to model and optimize the design of a ping-pong ball launcher. The system is similar to a grandfather clock, powered by a weight with an escapement, pendulum and cams controlling the timing. The exciting thing about this project was that we were able go from design to optimization and then to building and demonstrating. Below is a drawing labeling some of the design parameters and a picture of the final product.
1. Prototype haptic grippers for laparoscopic surgical teleoperation – 2006-2007
(worked on the project in a team with Will Howison and Canek Phillips)
Designed, built and tested a master and slave gripper teleoperation devices (shown below, master on left, slave on right) that emulate the function of laparoscopic surgical tools. The prototype has used in experiments testing the effectiveness of haptic feedback in surgical teleoperation devices.
Howison, W., R. Winck, C. Phillips, N. Xiao, N. Kishan, J. Araujo, S. Kreml, and M. O’Malley, “Haptic Grippers for Surgical Teleoperation Research,” 24th Annual Conference on Biomedical Engineering Research, February 8-9, 2007.
Howison, W. and R. Winck. “Haptic Device Grippers for Surgical Teleoperation,” US-2009-0287351, published 19 November, 2009.
2. A haptic device for stroke rehabilitation – 2006
In work for Professor Marcia O’Malley, I helped modify a device that had been used in the lab for haptics research so that it could be used as a stroke rehabilitation tool. It was originally an elbow/forearm/wrist haptic device and it was modified to be only forearm and wrist but to attach to a Puma arm to provide full arm feedback. I helped modify the mechanical and electrical systems, but the largest part of the project for me was in coding a state machine so that a physical therapist could use the device without having to interact with the software. It was an excellent undergraduate research project because I got to do a little of everything. Below is a picture of the device, called the Rice Wrist, attached to the Puma arm where it would be used for rehabilitation research. Yes that’s me strapped in. Always test with the undergrad first.
3. Carnival track game – 2006
(worked on the project in a team with Brian Van Osdol and Javad Sanati)
This fun and exciting carnival game pits two players in head to head running-in-place competition. As the opponents run, their motion is sensed and monitored to actuate a motor which moves a character across a pulley system. Other features include stop/reset switches and the ability to give players handicaps. The software was written in LabView using an NI Elvis board for A/D and D/A. So, who will win, the Jetsons or the Flintstones? Only your feet can decide!
4. Remote handling equipment for the spallation neutron source – 2005
During an internship with the Robotics Group at Oak Ridge National Lab I designed and oversaw the machining of parts to be used in remote handling equipment at the Spallation Neutron Source at ORNL. A few of the parts are shown below.