Move research-interests and Skype Telepresence Rover pages from google sites

projects/index.md

@@ -0,0 +1,16 @@
+<!--
+title: Projects
+-->
+# Projects
+
+## Personal Projects
+- [Skype Telepresence Rover](./skype-telepresence-rover/skype-telepresence-rover)
+- Dyna Myte 2400
+- Javascript CNC Machine Emulator
+- Ultimate Word-of-the-Day
+- K'nex Adding Machine
+
+## Employment
+- Co-op Experience
+- ATA Customer License Manager
+- Enhancing TIE

projects/skype-telepresence-rover/index.md

@@ -0,0 +1,32 @@
+<!--
+title: Skype Telepresence Rover
+-->
+# Skype Telepresence Rover
+
+![The current prototype](IMG_20110109_184027.jpg)
+<br>*The current prototype.*
+
+While on co-op in California, I often used Skype to video call my family. I even attended my brother's high school graduation virtually via Skype; from a laptop, I chatted with all my old friends from high school and joined in group conversations. People carried me around so I could say hi to everyone at the party. It was a great experience, but I depended on others for mobility. At one point, I was set down facing a wall for 10 minutes until someone noticed and turned me around.
+
+This got me thinking, what would it take to have a real telepresence? A full-blown telepresence robot like an [Anybot](http://anybots.com/) or a [Texai](http://www.willowgarage.com/pages/texai/overview) was out of the question; I'm a college student on a shoe-string budget. But with the shrinking cost of electronics, I was convinced I could make a telepresence robot on a shoestring budget. All I had to do was integrate some existing technology: Skype, a netbook, and an RC car. I estimated I could make a telepresence robot for less than $300!
+
+Part List:
+- A netbook with 2GB RAM, acquired from Craigslist for $150
+- A $60 RC car from Meijer
+- An $30 Arduino microcontroller to connect the laptop to the RC car
+- Some wood and screws to mount the laptop and Arduino to the RC car
+- Python code (using the Skype API) to send and receive steering and speed commands
+
+[![Arduino + H-bridge test](http://img.youtube.com/vi/QsiM0Cjdzhk/0.jpg)](https://www.youtube.com/watch?v=QsiM0Cjdzhk)
+<br>*This is from an earlier prototype that was built on a chassis that did not have a built-in motor controller.*
+
+[![Skype Rover 2 - H-bridge Hijack](http://img.youtube.com/vi/UroWCdb6SUo/0.jpg)](https://www.youtube.com/watch?v=UroWCdb6SUo)
+<br>*For my second prototype, I used a complete electric car chassis, and reverse engineered the built-in motor controller. This video demonstrates applying a 3-5V input causing the wheels to spin, drawing power from the built-in battery, not the voltage source.*
+
+I had hoped to make a video of the robot in action, but have been too distracted to add the joystick control needed to steer the robot properly. In the meanwhile, above are a couple of videos I took during the development of the Skype Rover.
+
+The idea is really simple. The rover has it's own Skype Account. When you call it, it automatically accepts the call and puts your video in full-screen mode so others can see you well. On the controlling end, a Python code lets you send steering and speed commands using the mouse or a joystick, and on the rover end is Python code that receives the commands and sends them through a USB cable to the Arduino, which controls voltages wired into the RC car's motor controller. It is a masterpiece of integration, really. The technologies for telepresence were just there, waiting to be connected with a little bit of solder and Python by some clever individual. If it works, thousands of DIY hobbiests could build one of these things in a weekend.
+
+The project is about 90% complete. Upon my return to UC, I decided I wanted to be able to give tours of the UC Engineering building via the Skype Rover. However, initial tests revealed the engineering building has too many Wi-Fi deadzones to be able to drive the robot down a hallway without disconnecting. I also discovered that the open-loop control combined with latency issues made the rover extremely difficult to drive. Disheartened, I have temporarily shelved the Skype Telepresence Rover in favor of pursuing other projects such as learning ROS, the Robot Operating System from Willow Garage.
+
+After finishing most of the code, I searched for similar projects on the Internet and found [Sparky Jr](http://sparkyjr.ning.com/). Sparky Jr. has a similar goal; however, my rover was less expensive because of the decision to use an RC car chassis and a netbook rather than a Roomba chassis and a Mac mini.

research-interests.md

@@ -0,0 +1,49 @@
+<!--
+title: Research Interests
+-->
+
+# Research Interests
+
+## Past Research:
+I was selected for an NSF-funded undergraduate research experience at the University of Cincinnati. I and two other students researched novel stereovision filtering, computer vision, and obstacle avoidance algorithms on UC's intelligent ground vehicles.
+
+As a graduate student at Drexel University, I participated in the DARPA Robotics Challenge and developed experience with the Hubo humanoid robot.
+
+## Past Areas of interest
+### Human-robot interaction
+How can we ensure our robot-filled future is full of perceptive, helpful, friendly robots, rather than dumb, pesky, frustrating ones?
+#### Why I'm no longer interested
+As of 2012-2014, the technology is just not mature/reliable enough to do meaningful experiments with humanoid robots.
+The robots I had to work with were too dumb, pesky, frustrating for even *me* to put up with. And that's saying something.
+
+### Perception
+How to make sense of the wealth of sensor data available to robots
+#### Why I'm no longer interested
+Actually, it would still probably be a pretty interesting field. But frankly, it looks like
+machine vision, 3D reconstruction from video, augmented reality physics, etc all seem
+to be coming along just fine.
+
+### Machine learning
+How to organize and apply knowledge
+#### Why I'm no longer interested
+I'm a machine learning hipster. I was into it before it was a buzzword.
+After studying it for three years at university, I was kind of disappointed in a "this is it?" way.
+Then deep neural nets took off, and the tools became so plug-and-play that everyone
+and their dog can build an "app" that uses machine learning. While neural nets do indeed organize
+knowledge, to date the ML field has been obsessed with results/performance,
+without concerning itself with *understanding* how these nets are organizing that knowledge.
+It is a very shallow victory.
+
+### Machine ethics
+What do we tell cognitive agents about how to behave? Can we define or enforce ethical robot / AI behavior?
+#### Why I'm no longer interested
+I don't believe it matters, because they can't be any worse than humans.
+
+
+## Continued areas of interest
+### Cheap hardware revolution
+How can we best take advantage of game-changing cheap hardware, like the Kinect and Android phones?
+### Ubiquitous Computing
+As the number of embedded computers increases, how can we keep the interface complexity from also increasing?
+### Disposable Computing
+How cheap can computing get? Raspberry Pi Zero, unlimited free cloud deployments... could we start replacing our smartphones every week? Every day?