/* ESPOSITO ADDED META TAGS 6/30/08 */

Wednesday, February 25

Sail Bot

Imagine trying to teach a robot to sail. That is exactly what this group of students had to do when they entered the first international autonomous sailboat competition, hosted here on the Chesapeake Bay. They retro-fitted a standard, small, sailboat with digital sensors such as GPS, a compass and wind speed indicators. All that information was fed into an on-board computer which controlled various motors to adjust the rudder and sails.

Our team, which consisted of systems majors and naval architects, placed second in this international competition!


Tuesday, February 17

Roombas

Most of us have seen the iRobot Roomba vacum cleaner robot. Professor Esposito wrote a Matlab Toolbox for the Roomba that allows you to control a Roomba from your PC.

The ES451 Mobile Robotics class used this along with a pretty sophisticated scanning laser rangefinders to program the robots to track along within 2 feet of the wall (the tape lines on the ground are just for reference the robot cannot see them) and when it finds an open door it is able to drive through it.


Mids Warren Leonne

These are the same sensors that have been used succesffully in the DARPA Urban Challenge

Tuesday, February 10

Alumni


Even wonder what some of our Alumni do when they enter the civilian engineering world?

Dear Dr. Knowles,

My name is Matt Wallace. I am a 1984 USNA grad, and you were my advisor and professor a long long (long) time ago.

I was a systems major, and was at the time interested in robotics. Of course, I had little opportunity to exercise that interest while driving fast attack subs around various oceans - though that was mighty cool too. However, when I left the Navy, you were kind enough to write me a recommendation for graduate school in 1990 - despite the fact that I was far from your best student. I ended up at Caltech for a masters program, which I somehow survived (way harder than plebe summer), and was then hired at NASA's Jet Propulsion Laboratory, which is run by Caltech.

I believe it's been 20 years since you have heard from me. And no, I don't need a recommendation for a PhD program. In fact, I really only wanted to send this to let you know that your support made a difference for me. I have been working in planetary robotics now for most of those 20 years.

Over the last 15 years or so, I've been focused on the Mars Program run here at JPL. I played a part in the 1997 Mars Pathfinder mission (I lead the power system development on the small Sojourner rover among other things). I was also the assembly test and launch operations (ATLO) manager for the 2004 Spirit and Opportunity rovers, as well as the Opportunity mission manager for the Mars surface phase. And I am now the spacecraft manager for the 2009
Mars Science Laboratory (MSL) mission. MSL is a beast - 4000 kg all together with a 900 kg car-sized Rover stuffed with 10 different science payloads.

As you might expect, there is not a lot of exotic cutting-edge technology on these vehicles since we focus on high-reliability, mature approaches were possible. You're students today would probably scoff at our Power PC computers and 1553 bus. But the systems engineering and integrated design requirements push the boundaries of the industry. And of course, the application-level is fairly unique (not a lot of folks trying to land on Mars).

I still draw on the advice you gave me 25 years ago - I learned a lot about how to solve problems from you I believe. And I would not have had this career opportunity at JPL without the sequence set in motion by your recommendation.

Matt Wallace, Systems Engineering, Class of 1984, now at JPL

Sunday, February 1

Trackin'

Tracking moving objects is an important objective for many autonomous systems. Computer vision is one way to do that.

Anthony Guinipero
Using cues about color and shape helps you find the object in the image. But when the object moves fast or there is shadows or glare the results can be disapointing. Enter the Kalman Filter...

Lisa Knauft

It helps blend noisy measurements with a predicative model of motion resulting in the "smooth like butter" trackers seen in these videos.

Stephen Allen