Bartley Gillan

Bartley Gillan

I am currently a MSE Robotics candidate at the University of Pennsylvania affiliated with the GRASP Laboratory. I completed my BS in Electrical Engineering from The University of Texas at Austin in 2008. In short, I like to build things. This page serves as a portfolio of different projects I have completed over the years. I can be reached at .

Vision-based Mobile Robot Navigation
Fall 2008

This two-wheeled robot was programmed in MATLAB to navigate from a color webcam. Three separate subtasks were accomplished with this robot platform. The simplest task was to drive towards a red object and stop as close as possible to a certain distance away from the object. Our group accomplished this with one of the most minimal distance errors. The next task was for the robot to drive in a “figure eight” with minimal path deviation. My group's solution was to use mirrors in front of a mounted camera to track two LED beacons marking the circle centers of the “figure eight” path. We were the only group to use a challenging robot-mounted camera approach. The final task was to navigate the robot through three sets of red gates in a “slalom” fashion as fast as possible. Our team acheived the fastest time and won 1st place in the class competition.

Supervised Learning Classification in LOST
Fall 2008

This project's purpose was to correctly classify the identity of face images and one second voice audio clips as one of five characters from the TV series, LOST. Our group used a combination of machine learning algorithms such as neural networks, support vector machines, hidden markov models, and k-nearest neighbors. I focused on implementing hidden markov models for the voice audio classification task. Our methods resulted in an image error of 0.273 and an audio error of 0.4, which gave us 6th place out of 30 groups in the machine learning class.

Robocup Soccer Simulation
Spring 2008

I took a course on autonomous multiagent systems my last semester at The University of Texas where the experimental domain was the Robocup soccer simulator. Each individual developed a team to compete in a class tournament held at the end of the semester. My team placed 2nd in the class tournament. I leveraged the UVA Trilearn 2003 Robocup team's C++ code base and implemented some ideas from the ATT-CMUnited-2000 Robocup team to build an aggressive, offensive-focused team. This strategy paid off in scoring many goals against opponents and went undefeated into the final match. My final report can be found here.

Mobile Ad-hoc Network Open Robot Platform
Fall 2007

This robot was my award winning undergraduate senior design project at The University of Texas. My advisor was Dr. Sriram Vishwanath and the project is associated with the WNCG at UT. The robot was designed from the ground up to be modular, open, and easily replicated. This was the first of a planned swarm of mobile robots to function as a testbed for research into wireless algorithms on mobile ad-hoc networks. The motivation for this project was to one day be able to have the ability to send out a swarm of autonomous robots to survey and gather data through sensor packages in specific areas of interest. The robot was semi-autonomous where it could randomly explore and visually survey an area avoiding obstacles on its own, or it could be remotely controlled from a website. It was built upon a RC car base and was equipped with an x86 mainboard running OpenEmbedded Linux, ultrasonic range finders, WiFi, and a webcam for video recording. The project was awarded 1st place out of all electrical engineering senior design projects that semester. See this paper for more details.

DARPA Urban Challenge
2007

I joined the Austin Robot Technology team at the beginning of 2007 along with The University of Texas Computer Science department to help develop a viable entry in the DARPA Urban Challenge. The team consisted of the existing ART team and a group of undergraduates led by Professor Peter Stone and PhD student Patrick Beeson. Most of the ART team focused on the physical hardware of the robot, Marvin, while much of the software was written by the UT group. Early on we performed well enough to make it out to Victorville, CA for the National Qualification Event where we advanced to the semi-finalist round to be among 21 remaining teams before the final competitors were selected. My contributions were mainly in the areas of high-level navigation and perception code as well as testing. An article I coauthored detailing our approach was published in the Journal of Physical Agents here.

Lego Mindstorms “Capture the Flag” Robot
Fall 2003

The first robot I ever built was of the Lego Mindstorms variety for an Intro to Electrical & Computer Engineering course. There was a class competition set up where each team's robot would go head-to-head against another to “capture the flag” of the other team by physically moving into their base first. The robots had only bump sensors and an optical sensor to detect and follow a line. My team's dependable robot took 1st place in the competition.

last updated: 11/05/2009

	Vision-based Mobile Robot Navigation Fall 2008 This two-wheeled robot was programmed in MATLAB to navigate from a color webcam. Three separate subtasks were accomplished with this robot platform. The simplest task was to drive towards a red object and stop as close as possible to a certain distance away from the object. Our group accomplished this with one of the most minimal distance errors. The next task was for the robot to drive in a “figure eight” with minimal path deviation. My group's solution was to use mirrors in front of a mounted camera to track two LED beacons marking the circle centers of the “figure eight” path. We were the only group to use a challenging robot-mounted camera approach. The final task was to navigate the robot through three sets of red gates in a “slalom” fashion as fast as possible. Our team acheived the fastest time and won 1st place in the class competition.
	Supervised Learning Classification in LOST Fall 2008 This project's purpose was to correctly classify the identity of face images and one second voice audio clips as one of five characters from the TV series, LOST. Our group used a combination of machine learning algorithms such as neural networks, support vector machines, hidden markov models, and k-nearest neighbors. I focused on implementing hidden markov models for the voice audio classification task. Our methods resulted in an image error of 0.273 and an audio error of 0.4, which gave us 6th place out of 30 groups in the machine learning class.
	Robocup Soccer Simulation Spring 2008 I took a course on autonomous multiagent systems my last semester at The University of Texas where the experimental domain was the Robocup soccer simulator. Each individual developed a team to compete in a class tournament held at the end of the semester. My team placed 2nd in the class tournament. I leveraged the UVA Trilearn 2003 Robocup team's C++ code base and implemented some ideas from the ATT-CMUnited-2000 Robocup team to build an aggressive, offensive-focused team. This strategy paid off in scoring many goals against opponents and went undefeated into the final match. My final report can be found here.
	Mobile Ad-hoc Network Open Robot Platform Fall 2007 This robot was my award winning undergraduate senior design project at The University of Texas. My advisor was Dr. Sriram Vishwanath and the project is associated with the WNCG at UT. The robot was designed from the ground up to be modular, open, and easily replicated. This was the first of a planned swarm of mobile robots to function as a testbed for research into wireless algorithms on mobile ad-hoc networks. The motivation for this project was to one day be able to have the ability to send out a swarm of autonomous robots to survey and gather data through sensor packages in specific areas of interest. The robot was semi-autonomous where it could randomly explore and visually survey an area avoiding obstacles on its own, or it could be remotely controlled from a website. It was built upon a RC car base and was equipped with an x86 mainboard running OpenEmbedded Linux, ultrasonic range finders, WiFi, and a webcam for video recording. The project was awarded 1st place out of all electrical engineering senior design projects that semester. See this paper for more details.
	DARPA Urban Challenge 2007 I joined the Austin Robot Technology team at the beginning of 2007 along with The University of Texas Computer Science department to help develop a viable entry in the DARPA Urban Challenge. The team consisted of the existing ART team and a group of undergraduates led by Professor Peter Stone and PhD student Patrick Beeson. Most of the ART team focused on the physical hardware of the robot, Marvin, while much of the software was written by the UT group. Early on we performed well enough to make it out to Victorville, CA for the National Qualification Event where we advanced to the semi-finalist round to be among 21 remaining teams before the final competitors were selected. My contributions were mainly in the areas of high-level navigation and perception code as well as testing. An article I coauthored detailing our approach was published in the Journal of Physical Agents here.
	Lego Mindstorms “Capture the Flag” Robot Fall 2003 The first robot I ever built was of the Lego Mindstorms variety for an Intro to Electrical & Computer Engineering course. There was a class competition set up where each team's robot would go head-to-head against another to “capture the flag” of the other team by physically moving into their base first. The robots had only bump sensors and an optical sensor to detect and follow a line. My team's dependable robot took 1st place in the competition.