Previous Year - History Home - Next Year
Members
- President: Gabe Brisson (ECE '98)
- President: Rich LaBarca ('98)
- Jonathan Hurst (Meche '01)
- Charlie Reverte (Meche '01)
- Peter Sand (CS '00)
- Michelle Ungerer (Meche '01)
Projects
Awards/etc
Other Information/Notes
(from attached PDF from Rich LaBarca)
Carnegie Mellon University Robotics Club
Year End Report
Richard S. LaBarca
May, 1998
The Year in Planning
The CMU Robotics Club faced a major dilemma in 1997-1998, with only one school year left until the two remaining people able to run the club, Rich LaBarca and Gabe Brisson, would graduate. The club’s membership had dwindled to just these two in the past year, due to a loss of focus of the club as a whole and lack of interest of incoming students. With this in mind, Rich and Gabe spent part of the summer of 1997 planning the club’s focus and strategy to gain more involvement. At the beginning of the school year, the following goals had been solidified:
Continuous Motion Frame-Walker
For most of his junior year, Gabe had been looking into transforming the Moose frame-walker idea into a walker that could keep itself in constant motion while switching frames. Since it would be statically stable, concentration could be put on control software and sensing and it was more likely that a working machine would be completed in a year. The general feasibility and basic design layout were completed that summer with input from Nick Vallidis, former RoboClub treasurer and now a graduate student at the University of North Carolina. It was decided that the Daedalus motors, now sitting around after the failure of Daedalus’ lower frame rotation joint, would be used to drive the robot, and would therefore be the major design constraint. This eventually led the frame walker to be much heavier and larger than originally planned.
Because of the success of Gabe’s walking simulator/scripter for the club’s previous robot Sidewinder, it was decided that a simulator be built for the robot so that programmers could develop control algorithms, and new people could gain a better understanding of what the final robot configuration would be like. It was also decided that new people would be allowed to have input into the mechanical design, though the final goals of the robot were fixed and not up for discussion. This was to avoid long meetings of useless talk and redesign. To verify the validity of the design, a design review was to be held as early as possible, to ask members of the RI faculty to review and critique the design.
Vision System
During the 1996-1997 school year, when the RoboClub was floundering, Rich had been looking into developing a vision system for future SAE Walking Machine Decathlon entries. By the end of his junior year, he had developed a black and white system that could pick out a ball reliably. However, in 1998, the ball and cone, the objects to be located in several events, would be distinctly colored. This lent itself directly to color vision, so it was decided that a new color vision system would be developed during the year. Ideally, this system would be able to tell the position of a ball and cone, as well as correct dead reckoning inaccuracies using lines present on the Walking Machine Decathlon playing field.
Biped
At the end of his junior year, Rich proposed a Robotic Walking senior thesis to the Computer Science Department that was accepted and encouraged by John Wiss, the RoboClub’s academic advisor. It was to be done with John Burroughs, an ECE senior, also working on the LRI Nomad project at the time. Rich designed the initial software goals during the summer, and began coding some parts of it. Andrew Moore showed interest in advising, and though uninvolved for the majority of the development, became the faculty contact for the project. The goals for this project were to develop all the hardware and electronics for the robot, and for Rich to develop a software system to be used in experimenting with different control methods as his senior thesis. Therefore, at the end of the school year, the goal was to have a fully functional robot, and a software environment in which to play with it.
The Year’s Key Points
Membership
The 1997-1998 school year saw more useful participation than either Gabe or Rich could have imagined. A surprisingly successful Activities Fair recruiting process accomplished this, along with an influx of freshman interested in Robotics. Though approximately ten new people attempted to remain active, four made it through the entire year. This number is excellent, compared to the one or two person average the club is known for. Another plus is that the skills of the new members were not restricted to software end electronics, as is usual for RoboClub members. A table of consistently active members follows:
| Name | Specialty | Year |
|---|---|---|
| Gabe Brisson | Electronics/Low-Level | Senior |
| Jonathan Hurst | Mechanics | Freshman |
| Rich LaBarca | Software | Senior |
| Charlie Reverte | Mechanics | Freshman |
| Peter Sand | Software | Sophomore |
| Michelle Ungerer | Mechanics | Freshman |
Though each member had more skill in a specific area, many people accomplished tasks in other areas as well. Most importantly, each new member was more self-motivated than new members in previous years.
The biped project lost John Burroughs to his own personal issues, but gained heavy involvement by Jonathan Hurst, a freshman Mechanical Engineering major, and a Brian Olson, a freshman Computer Science major. Brian did not actively participate in RoboClub, but joined Jonathan and Rich in building the biped as part of a $1040 SURG grant obtained in January 1998.
Design Review
The design review was organized to get expert advice on the mechanical design of the robot, and to give us a deadline for pinning down the mechanical design. The attendees were as follows:
- Ben Brown
- Ralph Hollis
- Eric Rollins
- Mark Sibenac
- John Wiss
The reviewers gave us good advice on considerations for choosing bushings and bearings and pointed out problem areas we hadn’t considered. It proved to be extremely useful, and set a precedent for future design reviews.
Machine Shop Access
At the beginning of 1998, the RoboClub was able to finalize their access to the FRC machine shop. Colin Piepgras and Abe Crowley interviewed Charlie and Jonathan about their previous machining experience, and decided to allow Jonathan to use the mill and lathe, and Charlie to use the lathe. They discussed safety issues and explained proper cleanup and machine maintenance. All club members were allowed to use the shop tools excluding the mill and the lathe, including the band saw, cutoff saw, drill press, grinder, sander, and all of the hand tools.
One week before we planned to leave for the walking machine competition, the shop was packed away and moved to building E. The move was very inconvenient since the club was working intensely at this point to get the robot built for the competition. Even with the help of Matt Mason and the Manipulation Lab’s machines, building was brought to a slowdown. In a brief talk with Red Whittaker before we left for the competition, Red explained to Rich that RoboClub needs to make their wishes known more explicitly to avoid similar complications in the future. This was most certainly noted.
RI Contacts
Through the school year, increasingly solid communication channels were established between club members and FRC people. Most noticeable of these was an effort on the part of Rich and Gabe to keep Red informed, since he has been consistently helping RoboClub behind the scenes. In the course of several meetings, Red agreed on a set yearly budget for the club on the order of $3,000, and helped in the purchase of a new Gateway laptop and the allocation of a new Intel desktop. Oz was also a key participant in allocating money to the club, and helping with the purchase of the new laptop.
Due to the large amount of time Charlie and Jonathan spent in the shop through the year, communications channels have been opened to Colin and Abe, as well as various other mechanical people in FRC. Now that the shop has moved to building E, interaction will surely arise between RoboClub and LRI people.
The amount of help from RoboClub alumni has been substantial, with the most active participant being Ryan Miller. Ryan took a large part in working out the electronics in the walking machine, as well as helped in obtaining batteries and various other supplies. Charles Ouyang helped a lot with the initial mechanical design. Mark Sibenac has also been a constant source of answers for embedded questions and obtaining equipment, as well as participating in the walking machine design review. Nick Vallidis has kept his interest in the club and continues to encourage us from UNC.
Rich’s position in the MRCAS group has helped to gain the assistance of Mike Blackwell in various tasks. Mike helped us obtain an obscene amount of batteries from an old Navlab UPS and was a constant source of advice and encouragement.
Gabe’s old position in the Manipulation Lab and Matt Mason’s generosity gave us access to Manipulation Lab machines when the FRC shop was being moved.
In the future, several FRC connections can be established with the new leaders. Jonathan is a member of Ralph Hollis’ Microdynamic Systems Lab, and Hagen Schempf has offered positions to several RoboClub people. Next year, Gabe will be a RoboGrad, and Rich will be full time with MRCAS, and is attempting to become a RoboMaster’s student. Both plan to be active participants in the club.
Accomplishments
Jim
The walking machine was given the name Jim because no one had a better suggestion. Though Jim’s state at the end of the year was non-functional, it is considered a success due to the amount of useful, reusable work that went into its creation and to the fact that little work is required to turn it into a sturdy, powerful, machine.
The kinematic simulator used to develop control algorithms for Jim was a total success, and allowed new members to actively participate, even though they were not up to speed with software development for mobile robots. Most of this new software can be adapted to new robots.
The Jim control software was also a successful, though never used in competition. It could be run with or without the robot connected and provided a convenient environment to test event code and other functions.
A description of all aspects of Jim can be found in the SAE Technical Paper, Jim: An Intelligent Continuous-Motion Frame-Walking Robot, available from us. (We’re wondering if this could become a RI Tech Report.)
The Walking Machine Decathlon was a failure, in that Jim was not ready to compete by the time we arrived. The effort put into it, however, was quite large and an accomplishment unto itself.
The day we were supposed to arrive, after a few nights of sleepless work, our van broke down in the middle of Orson, Indiana. After discussions with Police and interesting rides with a gas station worker and a tow-truck operator, we ended up in a small town in Orson, behind an auto repair garage, next to a barn. For seven hours, we used an extension cord run from the garage to power a soldering iron and drill and managed to complete construction of the robot. We had neither machines, nor a place to work but we built Jim in the dirt using the van and Jim itself as workbenches. A local hardware store provided us with some supplies, and the local fish and chips diner provided us with food, but by the time our new van arrived, the amount of work done surprised us.
The competition itself was disappointing regardless of the fact that we didn’t have a working robot. Originality amongst the entries was severely lacking, and it felt as if we were penalized for having it. The members of other teams proved to be extremely friendly, and we had some good discussions while we were franticly trying to fix Jim.
Though the other teams were friendly and unpretentious, the organizers of the event were not. Though such details do not belong in this report, the amount of rudeness and general snobbery by Peter Nagy, the organizer at Northern Illinois University, and the general non-caring attitude of SAE officials, surprised and disheartened us.
As of the end of the year, Jim’s functionality is limited. This is due in part to the unavailability of machines at the end of the year, and complications in the design, including the excessive weight of the Daedalus motors. Its electronics have some bugs that need to be worked out, and the top frame translation mechanism is not properly built, but with a little more work Jim will be fully functional. The top frame, along with some other excessively heavy parts will be redesigned at the beginning of next year, and the electronics will be fixed, so that software testing can begin.
At the same time as Jim is being completed, construction on Jim2 will begin. The goal of this robot is to optimize the continuous motion idea without the restriction of the Daedalus motors and with new knowledge gained from the development of Jim.
Vision
The vision system is more successful than anticipated. All of the planned functionality has been achieved, with accuracy in the centimeter and sub-centimeter range. With the completion of Jim at the beginning of next year, vision testing can begin. It is also planned to create a new embedded, hardware and software modular vision system, incorporating algorithms from this vision system as well as statistical methods for object location. This system will be isolated in an embedded computer system, with only a uniform serial connection as the communication between it and a robot.
Iria
The biped, named Iria, is fully functional mechanically and electrically, though servo vibration and accelerometer accuracy are concerns that have to be addressed. The software system is developed to a beta version, but the full experiment-ready version will be completed in the summer of 1998. A stripped-down version of the software is used to control the robot for demonstration purposes and for hardware and electronic debugging.
Conclusion
A formal proposal for the 1998-1999 RoboClub year will be prepared over the summer and presented in September. This will include points discussed here, described in detail, along with budget and management plans.
This year was a success in that RoboClub now has successors for leading the club. Though Rich and Gabe will remain active in the coming years, the new group is more than able to produce excellent results and push new boundaries in the club’s abilities. None of the robots created this year will be left to rust in a corner, but will be further developed and redesigned until their design intent is satisfied.