Projects

Name	Description	Year
6-Legged Robot	My first 'robot', a remote-controlled 6-legged walking aluminum frame that I built based on plans from Gordon McComb's "Robot Builder's Bonanza." It had no sensors or control outside of a switch box I used to control the motors. If you didn't watch the leg positions, the robot would turn itself on end.	1995
3 Wheeled Robot	I upgraded the above robot frame to use 2 powered wheels, one castor, added a Motorola 68HC11-based microcontroller board called the "Handyboard," some bumper sensors, an IR proximity sensor, and a Pan/Tilt head with two 'eyes' composed of 8 photoresistors each. The Handyboard was programmable in C, so I learned to program on this robot - it would avoid walls, and the pan-tilt head could track a light source around the room.	1997
Revised 3 Wheeled Robot	I changed the frame of the above robot again, I gave it bicycle tires and put the third wheel on an arm that could be extended and pulled in, like an inverted R2D2. I also added sonar and infrared to the Pan-Tilt Head.	1998
Small 2-Wheeled Robot	This was a small 2-wheeled robot that used 2 servos modified for continuous rotation. A quick and dirty robot for a school project.	1999
High School Lego Competition Robot	My high school robotics club participated in a Lego robotics competition where the challenge was to build a factory automation robot that did 3 axes of manipulation, using only 2 motors. We built a little directional transmission to split the drive of one of the motors, and used the other for a grabber.	2000
EyeBot	My first robot with the UMR Robotics Club, which I helped found. This was a little robot that was an EyeBot controller built onto the chassis of a toy car. The EyeBot controller was programmable in C and had a built-in frame grabber and libraries for on-board computer vision methods.	2001
FIRST Competition Robot #1 'Lazybot'	The first competition robot for the newly renamed UMR Robotics Competition Team. The F.I.R.S.T. Competition (For Inspiration and Recognition of Science and Technology) which paired colleges and businesses with high school robotics teams to build a remote-controlled robot that participates in a non-violent challenge competition. This year's challenge involved pushing different sized kick balls into goals.	2002
FIRST Competition Robot #2 'Son of Lazybot'	This year's FIRST game was to be able to manipulate and stack plastic containers. Our robot was built like a fork-lift. We placed 6th in our regional competition this year. We also built in some simple autonomy and sensors.	2003
FIRST Competition Robot #3 'Lazybot's Revenge'	This year's FIRST challenge was to manipulate soccer balls, and at the end of a round, be able to do a pull-up on a bar in the middle of the field.	2004
IEI Hexapod	I worked on this robot at Imagination Engines Inc. The robot was base on Lynxmotion's Hexapod II frame. The project was to use sonar and accelerometers to build a self-reinforcing neural network architecture to control the gait of the robot. The project was moderately successful, the robot was able to develop gaits that allowed it to move forward, backward and turn in place, but the system had the most trouble with efficient gates. It would constantly find ways to trick the software into reinforcing the network inappropriately. For example, the robot would point the sonar at the ground almost immediately because the returns looked like it was moving toward a border. For the test shown in the picture, we cheated by rigging the robot so that it would always stay level, but we later had it working without supports using accelerometers at the head and tail.	2004
rho-Bot	The UMR Robotics Competition Team's first competition robot for the Intelligent Ground Vehicle Competition (IGVC). We had a very ambitious design for a large 3-wheeled omni directional robot that used a sonar array and stereo vision to navigate the competition course. Unfortunately we were not able to get the robot to move before the competition.	2004
I-Beam Reactor Patrol Robot	As my senior design project I built a robot to patrol the experimental nuclear reactor on UMR's campus. The reactor had a large I-Beam that ran the length of the building, over the reactor and the engineering bays. The robot was designed to hang from this I-Beam, carrying a camera and a pyroelectric sensor on a downward facing pan-tilt head, so that the robot could detect and localize anomalous heat sources in the building. The robot carried a 400Mhz Windows CE machine, with Wifi. You could control the operation of the robot from a base station or put it in autonomous mode where it would systematically scan the reactor. The robot was also equipped with a docking port so that it could run to one end of the I-beam to a docking station where it could recharge it's batteries, enabling maintenance-free operation.	2005
MouseBot	This cute little guy I built on the carcass of an old microsoft optical mouse. It still uses the original mouse sensor for dead reckoning, and I added motors and bumper-whiskers. He runs off of an Atmel 8051 microcontroller, has a serial radio, and can charge his NiMH battery through the whiskers and a contact on the bottom of the frame. He can find his charging station using an infrared photodiode mounted on the front.	2005
Optical Prime (OP)	OP is a re-vamped version of rho-Bot (see above) for the 2005 IGVC season. With a new computer, new AI algorithms, and completed mobility, Optical Prime was able to complete qualification at the 2005 IGVC and successfully complete 26 feet of the course before getting confused. In 2006 OP will be back with a vengeance!	2005
LAGR (Learning Applied to Ground Robotics)	LAGR is a DARPA-sponsored research project to develop adaptive learning methods for navigating ground robotics. The idea is to be able to drop this guy in the woods, give him a waypoint, and he'll navigate there autonomously. Additionally, if he gets stuck or encounters an obstacle, a learning algorithm helps him avoid similar situations in the future. Carnegie Mellon built the frame, but UMR students do the programming. LAGR uses 2 pairs of stereo cameras and GPS to navigate. All computation is performed with 3 on-board Pentium based small form factor computers. CPU for each pair of cameras, and one CPU for planning.	2006
Hector RN-1	Hector, my little man! Hector is based on the Robonova RN-1 Humanoid Robot kit from Hi-Tec. I upgraded him with a 3-axis accelerometer for balance, and a serial radio for more complex behaviors. I plan doing many hardware upgrades when I'm not so poor. For now, he has 16 digital servos and a microcontroller that is programmed in a BASIC variant.	2006
Tux-Bot	During the summer of '06 I was spending the summer in Seattle working for Boeing. A friend of mine was getting married, and I couldn't get back home, so my housemate built Tux-Bot here from parts around the house, most notably the Power-Wheels motors. I did the programming from Seattle. The idea was that I would teleconference into the reception, schmoozing through the interface. Unfortunately, I was busy on the day of the reception, and TuxBot was dismantled before seeing action.	2006
Draganflyer Robot Swarm	This is a swarm of computer-controlled Draganflyer rotocopters that are part of a research project I am working on with Boeing. The Draganflyers are part of a UAV testbed used to develop multi-vehicle control algorithms. I spent the '06 summer at Boeing developing some methods for swarms of robots so that they can maintain high-level behaviors under changing environmental conditions, and failure modes.	2006
Mini-Prime	Mini-Prime is a small model of the 3-wheeled locomotion scheme that Optical Prime uses. This allows the robotics team to develop and test motion algorithms on a small platform before tranferring it to the large, heavy robot.	2006