Brian Nave probes deeper into the design secrets of a past winner
by Brian Nave
Paul’s Robotics, a team led by college student, Paul Ventimiglia of Worcester Polytechnic Institute won the $500,000 first prize in the 2009 Regolith Excavation Challenge. 2009 was the third year for the Regolith Excavation Challenge in which teams had to design, build and operate a mobile robot that could dig up and deposit at least 150 kilograms of material from a simulated lunar surface and deposit it in a collection bin. None of the teams in the 2007 or 2008 competitions were able to meet the winning criteria, so following the 2008 event a suggestion was made to find a permanent facility for the box with the simulated lunar regolith. That home was found in the Research Park at the NASA Ames Research Center in Mountain View, CA. The 2009 competition was held in that facility which, since then, has been available year-round for testing of lunar devices. Paul’s Robotics literally left the competition in the lunar dust by collecting an astounding mass of 500 kilograms. The Regolith Excavation Challenge has grown up, gotten a bit more complicated, and is now called Lunabotics.
ROBOT recently caught up with Paul Ventimiglia and had a chance to reminisce about his design and what his win has meant to him and his team.
What was some of the biggest design challenges for your Regolith bot?
The required “signal delay”, which simulated the time it takes for signals to travel to the moon, drove all major design considerations for our robot, Moonraker. All teams had to connect to their control systems through an event computer which added a delay of 2 seconds for each signal transmitted from your robot, and 2 seconds from your controls to your robot. Some teams attempted to control their robots “live” and were forced to drive incredibly slowly as a result (~.5 ft/second). For our team, the delay made us think about a robot which could carry very large amounts of regolith all at once, up to 150kg, to lower the total number of times we would have to maneuver and line up with the collection bin. The bandwidth limitation of 1mbit/s was very restrictive for the amount of visual feedback we would have liked for controlling a robot remotely.
The biggest design challenge overall was the limited time. The amount of time it took our team to design, build, program, test, and deliver our robot was the exact amount of time we had, about 4 months with our schedules and lives. The time crunch added a very high level of stress which was continuously more challenging to overcome.
How did your team come up with your design? What specific design features are you proud of?
I was responsible for the mechanical design of the robot, and I drew inspiration from other competitors’ robots from the 2007 competition, and digging machines throughout history. I spent months searching the internet for examples of mining and earth moving machines, with “trenching” machines providing the most detailed inspiration. We competed in 2008 with a robot which looked very different than the 2009 version of Moonraker, but the digging scoops operated in the same way.
Jim Greenhaw was an original competitor in the 2007 Regolith Challenge, and his continuously digging scoops attached to roller chain were elegantly perfect. For the 2008 event, Jim added many more scoops closer together which would provide a nice smooth and continuous digging motion. I had no fear of running exposed roller chain in the harsh regolith material because I witnessed it working well for Jim’s robot. Therefore, the scoops were modeled for Moonraker mainly with inspiration from Jim’s robot, but I widened the scoops by about 3 times to increase the amount of material which could be excavated while driving.
The tank treads for the drivetrain were selected based on the proven track record of earth moving machines which operate in sand and mud all the time. Those machines have increased surface area to prevent the vehicles from sinking into the soft earth, so it was a smart choice for handling the regolith.
Most people say that they really like the simple way that only one actuator is used to both raise and lower the digging scoops in addition to dumping out the hopper. Using the single off the shelf electric ballscrew was an elegant approach for its reliability, repeatability, and ruggedness. I especially like the 2 constant force gas springs on either side of the hopper which are always trying to fight gravity and lift the hopper up with 500 lbs of vertical force. The ballscrew has to work against those springs when lowering an empty hopper, but they are very useful with a full load of regolith.
What made your design so superior to the others?
The biggest mechanical difference between our robot and the others was the wide width of our digging scoops, 30 inches, and the high speed at which the scoops moved, about 3ft/s. I feel that the wide cutting cross section allowed us to dig a huge amount of regolith without leaving deep trenches on the surface which could trap a robot. My philosophy on the fast moving scoops was simple; I felt that if each scoop moves quickly, than it only “sees” a small amount of regolith, so the reaction force could be lowered.
The main thing that made our robot perform much better than others though was the collective expertise of our team members. The team consisted of myself, Mike Ciaraldi, Marc Devidts, Jennifer Flynn, Brian Loveland, and Colleen Shaver. The design of the robot components and operation were discussed as a team regularly to ensure that it met our goals of being simple, and reliable. Redundancy was evident throughout the control system. The robot had 3 on board cameras, and a 4th located on the collection bin. Our drivetrain used both encoders and a gyro with the ability to switch which was used for keeping the robot driving straight or making accurate turns in case something broke.
Can you describe some of the challenges of operating a robot remotely with limited visibility?
The biggest challenge of operating the robot remotely was the signal delay. After telling your robot to move, it would be 4 seconds before you could receive any feedback that the robot actually received your command and reacted properly. We created an easy to use custom GUI and control system which was designed to allow the robot to execute many short autonomous routines. Even for driving, we would never be driving the robot “live”. We would use commands such as, “drive forward 3ft”, “turn 15 degrees clockwise”, “begin digging”, or “dock”. Our GUI allowed the commands to be clicked, and visually sent into a queue. With that method, we could queue up a series of commands, and the robot appeared to move seamlessly and continuously.
In fact, the NASA judges were so impressed watching over our shoulders that they thought somehow we were controlling the robot in real-time (even though we were of course connected directly through their signal delay computer system). We simply watched our camera feeds, and looked for problems or anomalies while the robot was performing its functions. Our GUI provided full telemetry of all sensors (encoders, gyro, potentiometers, limit switches, cameras) and real time updates of our continuous bandwidth usage and time remaining.
The most commonly asked question we get to this day is, “what are the lights for?”, referring to the 4 bright blue LEDs on the corners of the robot, and the 2 red LEDs on top of the digger. The answer is that they look cool of course, but more importantly, it is how we can see our robot through the immense dust cloud it creates. Moonraker is not shy about producing an excessive amount of dust, and we learned that in testing with our low quality sand in our test sandbox. My favorite view for driving involved using the camera mounted to the collection bin which provided a perspective view of the robot. After our first test in the sandbox simulating a competition run about 10 days before shipping our robot, I drove out to the auto-parts store and bought some LEDs that people add to their cars for accent lighting. We stuck them on the corners of the robot, and they aimed back and up at an angle, directly at our collection bin camera, and we could see the robot perfectly through a complete dust cloud.
Your team was the first ever winner of the Regolith Excavation Challenge event 3 years ago. Will you ever try again with Lunabotics? What are some of the most valuable lessons the team learned from that experience?
I would like to compete in the Lunabotics event, it appeared very well run, and has so many teams entering, and I love competition! I think about it a lot because there were so many things I would do differently if I had the chance to design another robot. You always learn a tremendous amount from building and testing something, and also from seeing other people’s creations. However, I have just graduated from WPI, so I cannot compete unless I become a graduate student… The most valuable lesson learned was that competition experience in general is invaluable. Our team has, collectively, decades of experience designing, building, and competing with complex robots. We knew the kinds of things to watch out for, which components to use or not use, and good methods for producing effective results.
The most valuable lesson learned was proving how important it was to precisely create the conditions we would experience at the competition. In fact, at home, I only ever once drove the robot while being physically in the same room, for the initial test in our sandbox. After that point, I only ever drove the robot from a separate room through closed doors, and for 30 minutes at a time, even if issues were discovered during testing.
What was your favorite part of the competition?
I have a great time traveling and hanging out with my teammates who are also my closest friends. We are always laughing and finding humor even in stressful situations. I think my favorite part of the competition was getting to meet the new teams we had been talking to online, and seeing veteran competitors again. We publicly shared every picture and video we had of the robot during our construction, because we did not begin building parts until August before the October event. We even had a live webcam in our shop often, and would chat with other teams while they watched us work.
Oh, and it was also pretty cool when they handed us the check for half a million dollars, with my parents and closest friends with me…
What types of engineering disciplines made up your team, and why did you choose these disciplines?
We are a varied bunch, mechanical, electrical, programming, networking, systems engineering, and of course robotics. Robotics engineering encompasses so many different engineering disciplines which have to come together for the final product. Our team drew upon all of our experience at various times in each area. All of us were experts at one thing, but we also all understood and were familiar with all topics involved, so we could seek advice from each other and make sure our ideas were logical.
Did any of your team members gain prior experience with robotics from participation in other competitions such as BattleBots, FIRST, VEX, BEST, etc…?
Myself and Marc have been competing in Battlebots since 2001. Colleen Shaver has been a student competing in FIRST robotics since 1996. I met Colleen and Brian (also a FIRST alum) in 2006 where they introduced me to FIRST robotics as we were all mentors of FRC Team 190 out of Worcester, MA. WPI provides the engineering mentorship for the Mass Academy of Math and Science. Mike is a professor in WPI’s unique Robotics Engineering department, so he is very experienced with all things robotic. Jenn was sort of new to robotics competitions, but very experienced at programming and working on projects with difficult constraints. Moonraker definitely has parts and common techniques used on both Battlebots and FIRST robots.
How has winning the Regolith Excavation Challenge changed you and your team member’s lives?
It has certainly changed my life for the better, and I am sure my teammates would agree. We got a lot of good publicity and recognition after the event which felt nice considering all of the months or years of work we had put in and risked. It opened up many opportunities for my career, because the project showed people a project which was challenging, exciting, and well executed. For me personally, it was an incredibly stressful and yet rewarding experience. I definitely felt burned out for a while afterwards because of how little sleep I was getting for months. I was also constantly worried about the competition, and hoping that I wasn’t wasting my teammate’s time by convincing them this was a worthwhile effort. Winning the event removed any guilt I had about wasting people’s time which was a relief.
What projects are you involved in these days?
Most recently, I have been involved in finishing up this little project called “college graduation”, and I am proud to say that I have just finished it and it was successful! I currently am working on a new robotic prosthetic hand which was the capstone project for my Liberal Arts and Engineering degree from WPI. The prosthetic hand has 6 DOF and is entirely self-contained, including battery and microcontroller. I will be continuing to work on that project outside of school as I just found out that Proto Labs has selected the prosthetic hand for their “Cool Idea!” award. I used Proto Labs’ sister company, First Cut, to machine the unusually contoured pieces which form the parts of the robotic hand and was amazed at how easily they could machine the complex parts, in one day!
Various projects are always in the pipeline, and I am excited about a potentially large project which I can only say would definitely draw upon my Battlebots experience. I am in the process of creating some products which will make the lives of hobbyist and professional robot builders easier. I am also working on finally finishing my bartending robot.
OK, OK… How did you spend the $500,000 grand prize?
I don’t think anybody on my team has done anything too crazy or lavish as far as I know… I have certainly bought more tools and robot parts with my share, but I have really been waiting and saving it until I want to invest in my future robot projects. I would certainly love if another Centennial Challenge came around that fit my team’s skill set.
Thanks for the opportunity to do this interview!
Our thanks to Morgan Berry and NASA JPL for providing many of the photos for this story.
|Paul Ventimiglia with Moonraker|
|Wall-E was a big hit with the kids!|
www.andymark.com, (765) 868-4779
Innovation First (IFI)
www.vexrobotics.com, (903) 453-0800
www.robotpower.com, (253) 843-2504