Ashley Guy

Doctoral Student
The Robotics, Biomechanics, and Dynamic Systems Laboratory
Department of Mechanical and Aerospace Engineering
The University of Texas at Arlington
ashley.guy@mavs.uta.edu




Doctor of Philosophy in Mechanical Engineering
University of Texas at Arlington
in progress

Bachelor of Science in Mechanical Engineering
University of Texas at Arlington

Bachelor of Science in Biology
Louisiana Tech University



Current Project

High Speed Forward Dynamics in Protein Models: Synthesis of Multiscale Method, Adaptive Coarse Graining, and a Modified Nose-Hoover Thermostat.

Nano-scale simulations of polymers require considerable time and resources to produce results. Two distinct approaches have been developed for reducing computation time and increasing the integration time step. The goal of this work is to show the synthesis of these two methods allows for additional savings. A modified thermostat is applied to control system temperature. Forward dynamics are calculated using Featherstone's Divide-and-Conquer Algorithm, capable of log(N) calculation times when parallelized [1].

Multiscale methods are useful for dynamic systems with large disparities among its components' time and length scales. Active forces in the model are scaled, bringing terms in the equations of motion to the same order. Calculation time is significantly reduced and time steps can be increased with minimal change in results. Previous work has applied the method to motor protein and estrogen models [2,3,4].

Adaptive coarse graining methods identify inactive joints and apply constraints. These constraints simplify forward dynamics calculations by treating inactive regions of the system as a single rigid body. Inactivity is measured by standard derivation of joint angles. Constraints are removed when reaction force exceed thresholds. This method has been highly successful in RNA models [5,6].

The Nose-Hoover Thermostat is a well-known temperature control mechanism. Kinetic energy is calculated and compared to a desired value, the error used in force feedback. Thermostat response is characterized by an on-target mean with short-terem oscillations. A derivative term was added to the control law and shown to be successful. Results also showed the potential energy of the modified system quickly decayed into a low-potential state, significantly faster than that of the original system. Large derivative and/or proportional gains trend with increased computation time [7].


[1] R. Featherstone. A divide-and-conquer articulated-body algorithm for parallel O(log (n)) calculation of rigid-body dynamics. Part 1: Basic algorithm; Part 2: Trees, loops, and accuracy. 1999.
[2] A. Bowling, M. Hagshenas-Jaryani. A multiscale modeling approach for biomolecular systems. 2015.
[3] M. Haghshenas-Jaryani, A. Bowling. Modeling flexibility in Myosin V using a multiscale articulated multi-rigid body approach. 2015.
[4] A. Palanki, A. Bowling. Dynamic model of estrogen docking using multiscale analysis. 2015.
[5] M. Poursina, K. S. Anderson. An extended divide-and-conquer algorithm for a generalized class of multibody constraints. 2013.
[6] M. Poursina, K. S. Anderson. Canonical ensemble simulation of biopolymers using a coarse-grained articulated generalized divide-and-conquer scheme. 2013.
[7] A. Guy, A. Bowling. Modification of Nose-Hoover thermostat to improve temperature response in molecular simulations. (In Review)


On The Shelf

Molecular Dynamics Simulation of SiO2 Nanoparticles in Molten Na/KNO3
Modified Nose-Hoover Thermostat

This work is part of a collaboration between our lab and Dr. Donghyun Shin, whose work includes investigation into the increased heat capacity of molten salts by inclusion of nanoparticles. It has been shown experimentally that, when the nanoparticles are present in the molten salt solution, microstructures form around the nanoparticles; it is theorized that the formation of these microstructures is responsible for changes in the measured thermal properties of the solution. The goal of this work is to show the formation of the experimentally observed microstructures using molecular dynamic simulations and verify the increased heat capacity using statistical mechanics calculations.

Intelligent Ground Vehicle Competition (IGVC)

The Intelligent Ground Vehicle Competition (IGVC) is an annual international competition for student teams to design and build a robotic platform capable of autonomously navigating an outdoor environment. The platform must travel through a series of GPS waypoints while staying within marked boundaries and avoiding obstacles; all of these tasks must be performed without any external localization or control.

I served as Mechanical Lead for the UTA IGVC team. I oversaw design and construction of two platforms: Marauder (2013) and Yoshimi (2014). These platforms used custom omni wheels allowing holonomic motion. The 2013 competition was the first time UTA had competed in over five years; our team placed 12th of 52 overall.

These projects involved chasis, suspension, and wheel design, 3D printing, resin casting, laser cutting and CNC milling. Significant manufacturing was done by team members on site. On-board sensors included GPS/IMU, Lidar, webcam and wheel encoders.

We were succesful thanks to the efforts of Dr. Chris McMurrough, Rommel Alonzo, Bardia Mojra and Scott Phan. We were warmly sponsored by Dr. Frank Lewis and the UTA Research Institute. Special thanks to Matt Middleton, Stephen Savoie, and Norman Spayd.

Mara

Yoshimi








SURF algorithm for identifying features
Project for CSE 5369
Robotic Vision: Sensing, Localization and Control

The semester project allowed me to write a Speeded Up Robust Features (SURF) algorithm for the 2014 IGVC contest. SURF identifies similar distinct regions between input data and a target image. Objects are identified by significant correlating matches. While the algorithm successfully identified the practice book cover, it failed to detect a traffic barrel. A lack of enough distinct features leads to inconclusive matching. Additional work could incorporate color-matching algorithms for improved accuracy.



Gyroscope Stabilization System for a Two-Wheeled Vehicle
Undergraduate Senior Design

Senior Design is a two-semester capstone course intended to give graduating seniors an experience much like what could be expected in industry. Students meet with faculty advisors and clients on a regular basis, but are ultimately autonomous and personally responsible for the satisfactory completion of the work.

I served as the team lead and control system engineer. Our team was responsible for designing and testing a method of stabilizing an enclosed, two-wheeled vehicle; this concept vehicle is similar in form to a motorcycle, but fully enclosed to provide safety features and creature comforts more akin to an automobile. A two-wheeled vehicle at rest behaves much like an inverted pendulum: unstable and requiring some method of active control to remain upright. After analyzing the dynamics of such a system, our team designed and built a proof-of-concept of a gyroscope disc on an actuated gimbal. Given angular momentum of the disc, as the gimbal axis is actuated, a precession torque is generated that can be utilized to keep the vehicle upright.






last updated May 13, 2015