Why I coach these girls . . .

Onwards and upwards!

Apologies for not having blogged in a really long time. We’ve made a lot of progress albeit some pretty interesting setbacks, which is expected in all programs of this sort. I definitely think that experiencing a setback, dealing with it emotionally and moving on is an excellent experience for anyone to have at an early age. One of the other mentors mentioned something – we all learned to walk when we were babies by falling down. Some of the girls are lucky enough to have gone through elementary and secondary school without having to really try, and have perhaps forgotten the feeling of learning to walk by falling down and picking oneself up. The first time one falls again comes as a hard thud and then we learn that it is part of growing again. I’m excited because I saw it happen in the last ten days and everyone is on a roll again. This is perhaps lesson that a smart young woman can learn from being part of this robotics team – I value it above learning the technical aspects of robot building.

Here’s where we are in terms of robot building . . .

Chassis: Hahaha. I was banging my head against a wall on Friday evening after Mark Siminoff (senior IDEO engineer and leader) came and pointed out that Jim and I needed to drive all the mecanum wheels independently. Jim and I had run ahead and set up the mecanum wheels as if they were regular wheels in tank steering style. Once I drew the free body diagram, I realized where we had gone astray. No big deal. We tested out the tank steering on Saturday afternoon with some simple button driven drive code that Caroline and Erin worked with me to write (press a button to determine direction and the amount of joystick tilt determines magnitude). The chassis sounds like the mouse that roared with the mecanum wheels!

As of Wednesday, we have the chassis hooked up for four wheel drive. We had an extra two gearboxes which we bought for backup, but are using now. Caroline, Sherri and Erin put together drive code that we almost got working today. I think we need to cut the fat from their code i.e. strip all the default code examples because something is overwriting their PWM outputs. I think they are close though. I’m excited to see them ‘close the loop’ with the system finally. Thanks to Jessa and Karlin for their support of the programming team!



Arm: Alas, Kerk Motion Products has not been nice to us. I accomodated Jimmy’s enthusiasm by working out a deal with Mcmaster to get a 24” teflon coated lead screw (four starts, two turns per inch). However, Kerk, even after Mcmaster requested that they ship overnight, shipped the damn lead screw via ground. So it is currently somewhere between New Jersey and Chicago on a truck. And I ordered it on January 22! In any case, we’ve purchased a backup acme precision lead screw and nut from Mcmaster that should arrive tomorrow to drive our arm.

Meanwhile, Nandini and Anne did a superb job of putting together an ACTUATED arm prototype which they built out of 80-20 and piano hinges. I am really impressed with their ability to make the leap into prototyping. This is the first time I’ve seen in four years a build team proactively hooking up an actuator to their mechanism and driving it. How cool. In the process, we learned that we need to gear up (speed up) the follower sprocket on the lead screw. The motor currently runs at 45 RPM = 0.75 revs per second. For 0.5 inches per rev, it takes us forever to go up 20 inches, 30 seconds precisely, which is like the majority of the 2.5 minute game. So, we’re planning to gear up at least by a factor of four (down to eight seconds of travel), if not a factor of six (five seconds of travel). We have plenty of torque (limited by the 150 lb load that the lead screw nut can take). Nandini and Anne worked out the lead screw problem on the whiteboard.



Today, Jim and the gals got their heads around some of the issues in building the tower to support the arm. A lot of it was about geometry, and part of it was about what to choose as a material. Do we use 1X1X1/8” wall thickness aluminum or 1/16” wall thickness? Jim put lengths of both material in a vice and hung 10 lb weights off the end of the lengths. Everyone got a chance to push down on the cantilever beam and develop an intuitive sense of bending stiffness and strength. This is what I mean by intuitive engineering!



Gripper: Kudos to Tiffany Card for working with the girls this weekend on ProE to get a better sense of a number of gripper mechanisms and methods of actuation. I think we’re still looking at both linear and rotation actuation as well as grippers that have appendages that move in parallel or pivot about a point.

Once we have the arm and the tower set up, I think we can build on top of it pretty easily. Certainly this is not necessary how I would lead an internal IDEO program, but having accepted and come to terms with the sculptural aspect of robot building, I’m pretty excited and have a lot of faith that the intuitive that the build team and the mentors have developed via PCV and foam core prototypes will really help to inform us in the final design as we ‘sculpt’ it.

Electronics: Big cheers to Doug Bourn for helping us swap the SMT resistors on the bad gear tooth sensor board. It was great for Erin, Neli, Ginna and Arushi to watch Doug use solder wick and pop those tiny resistors off the PCB and swap them. We did get both GTS sensors working and the group of us hooked everything up to a power supply and watched the pulse on the oscilloscope on Monday.



Neli and Ginna also learned how to strip wire and crimp on lugs and today, Ginna and Doug built a control box for switching in and out of autonomous. I think Neli and Ginna are well on their way to helping us put together a well organized electronics board this year (which we really need to do – our pile of wires is starting to get a little scary).



Anyhow, things are going well with 12 days left. This being my third time through it, I think the emotional rollercoaster is a lot more predictable now. I hope that the girls are learning this too!

Can’t wait until the lead screws are delivered tomorrow morning. Robotics is really my excuse to geek out, code, build stuff, and tinker without the heavy responsibility of running a program at work. Thanks to everyone for your patience, generosity, support and food!

Don’t forget to check out our photo gallery here:

http://picasaweb.google.com/castilleja.robotics

Let’s start building!

Apologies for not having reported back for some time. We made a lot of progress on the design front in the last two weeks. I’m always excited to visit Casti because the gals always have a new prototype to share when I get there. On Saturday, Jim talked about using the KerK lead screw and by Wednesday, the gals had finagled an 80-20 version of the arm with all the pivots and slides working!

The game plan for the next week: Build the chassis with Jim and reassemble the kit chassis so that programming can work in parallel. This means we have to:

* Go buy all the materials that we need for Jim to start putting stuff together with us. This means a trip to Alan Steel on Monday, January 28 @ 3:00 pm. Caroline is putting together a cut list for 1X1 and 1X2 6061T6 aluminum square tube stock.

* Getting together a list of stuff to buy from Mcmaster for delivery by Monday. This includes long axles, flanged bearings to fit into the meccanum wheels, extra master chain links, extra sprockets and any other fasteners we would need.

* Planning and getting any long lead items like the 24” Kerk Motion Lead Screw (BZ8050T) through Mcmaster. This should be arriving next Friday. Eeek.

* Taking off the new stuff from the kit chassis i.e. the gear boxes and putting on previous years stuff so that we can start building the new chassis. While waiting for the new chassis to be built, we can experiment with the meccanum wheels as well.

* Start building this year’s electronics board. We can put it on a piece of corrugated plastic again. We should have this ready to go to hook up to all the motors before the chassis is done!

* We gotta start looking at our weight! Check out photos of the motors and wheels on the scale on our picasa site.

Chassis: I think the gals have settled on building a four-wheel drive chassis with four meccanum wheels. This should be really exciting and interesting this year. We took a look at what the frame should look like this year given the stack up of the 3” thick meccanum wheels and the associated sprockets and fasteners.

Jessa and Doug did some serious thinking on the driving code on Wednesday which is cool. I think it’s time to start coding up driving.c and getting the interrupts working while the Ann, Nandini and the build team work together to get the meccanum wheels and the kit chassis assembled.



Arm: Looks like we can do a power screw drive on an arm (third class lever – effort between fulcrum and load), which should give us more precision and less backdriving than with the usual sprocket and chain. I am concerned that we’ll need to be able to fold the arm back up at the end of a match when we take the robot off the field. We’ll be using the Taigene van door motor, and likely doubling the output speed to 100 RPM (since we only travel 0.5 inches per revolution of the nut). The Kerk Motion products triangular nut limits us to a 150 lb load, which at an 80% efficiency leaves us with 120 lb. We can probably safely place the pivot point at ¼ the length of the arm (20 lb at tip X 4 = 80 lbs at pivot, not including all the inefficiencies).

Serious math this weekend going on with David and Tiffany. I think David taught an entire statics class in an afternoon to the kids with a foam core prototype of the linkage. And then Eric pitched in working with Tobi to figure out how much load the lead screw could support.





Gripper: A bunch of our initial prototypes out of PVC and foam core helped us to understand just how gianormous this part needs to be. We’re struggling to figure out how two or three pronged gripper is going to start inside the starting dimensions (28X38X60). It needs two extra layers of actuation just to do this (passive or active I do not know). I think our goal is to get the chassis and the arm up and running soon and then develop and attach the gripper separately at the end.



We have 26 days left (19 days in). I am definitely feeling the pressure to move towards building! Thanks to all the parents and mentors for their support!

Emily

The promise of variety

My favorite days mentoring are those that involve a little bit of everyone – something mechanical, something electrical and something involving programming. It reminds me of the complexity of the problem these girls are trying to solve and how much fun a multifaceted problem really is.

Jessa, Doug, David and I were there today. The teams have made some excellent progress on all accounts. Annie and Divya took my homework assignment seriously and actually built a small scale model of not only the overpass but of the mechanism they have been thinking about for lifting the ball out of popsicle sticks and a baseball to represent the game piece. These gals are smart. Real smart. I was also touched that Sherri and Caroline took it upon themselves to build a gripper out of PVC. I really do suspect that once we show them a way to prototype or a way to do things, the exposure is enough for them to take it and run with it. It’s getting over the hurdle of knowing what to do that is hard. I don’t think they suffer very much from what Bob Sutton calls the ‘knowing-doing’ gap.

For dinner entertainment, I pulled up a few sketches in Solidworks to try to explain four bar linkages to the kiddos as it applies to designing their lift. I emphasized that Solidworks is a tool to help them make decisions and I think that using the 2D sketching features would benefit them greatly in figuring out some of the geometries they are trying to build. Jimmy and I had a good working discussion today over lunch (as he was waiting to get through online to get burning man tickets nonetheless) and we’re interested in some sort of four-bar linkage or scissor lift. David brought up the extendable ladder mechanism which I think we can do easily with a winch. He also had a good point about those great drawer tracks you can get from home depot. Perhaps we can use drawer tracks along with one side of the scissor lift to give it some stability, especially near the load. More later on this.

While David coached Ann and Nandini on the prototype chassis assembly, Doug and I chatted with the gals about a few important electronics / programming topics including PWM and duty cycle, how moving the joystick translates to a motion in the chassis, and some ways to work around the over responsiveness of Scottie (last year’s robot). Driving Scottie, I suspect, is a bit like trying to fly a helicopter. Scottie was EXTREMELY sensitive because, to my awe and surprise, apparently the girls took the input signal from the joystick, CUBED it (took it to the power of 3 / talk about serious processing meltdown) and then divided it again by some number. Doug then had a great point about using a lookup table instead, so you are guaranteed with an output within your boundary conditions and a lot quicker processing. Doug and I also talked about using an accumulator and other time-based strategies to reduce noise in the signal.



We also sat down and wrote code on the board. Mentors who are interested in coding, please let me know and I will get you a copy of the compiler and default code. I can walk you through the default code (it is messy). Albeit the big family of c and h-files, the girls should only be concerned with modifying user_routines.c and user_routines_fast.c (for autonomous / hybrid period). In the context of programming an accumulator, we talked about global variables, constants and arrays. I actually think we should encourage them to work on categorizing their functions into library files like driving.c / driving.h, lift.c / lift.h and gripper.c / gripper.h so the components can be developed by separate coders and then incorporated into the program. Then default_routine (void) as a function can be kept very simple and clean. I would actually like one of us to strip user_routines.c down to the foundation so that all this extraneous code doesn’t overwhelm the kids.

This was definitely a good day. We have momentum. I know the kiddos are worried about not having designs locked down yet and feel behind, but heck, neither does my project at work! We’ll figure it out soon. I’ll be there Saturday. See everyone there!

End of Week One

This is really déjà-vu for me blogging again. We started up these blogs at the beginning of the 2006 season and they’ve morphed into and out of use and back into use again over the last two years. The purpose of these blogs are to keep the mounds of people who support this robotics teams either daily, weekly or monthly caught up on news so that we can all dive right in when we are together at Castilleja. Sherri has committed from the student team to blog regularly and I have committed from the mentor team to do the same.

After a tough first week, I think we’re on a roll. I do believe that the key to success mentoring this team is to get them hands on with problem solving and to let their intuition guide them once they start experimenting through this design phase. Fearless team leader Erin O’Malley just got back yesterday from a long journey to India, so give her a hug or high five when you see her! Here’s where we are . . .

OVERALL TEAM OPS: Running much more smoothly. It’s a really tough year because we have a lot of newbies and a lot of mentors. The girls who are leading want to get into the nitty gritty of designing but they have to manage us big kids AND the little kids and give us something to do. On Saturday afternoon, the mentors and the team leads had a meeting, and we decided that for the next two weeks, to try tackling the design with small teams focused on one part, rather than keeping everyone within their discipline (i.e. build, programming etc.).

CHASSIS / OPERATOR INTERFACE: (Caroline, Sherri, Allie, Anne, David (Mentor)) Lots of progress this weekend. The team is building a prototype chassis with the kit of parts with all the motors and electronics attached so that they can start working on the coding, which I think is a brilliant idea. They had a lot of driving issues with Scottie last year, a lot of which I think can be solved using smoothing algorithms (moving averages, PID controllers). Also, David mentioned that he’s used a yaw sensor to detect slipping so that we can also adjust for straight driving using electronics and code. We also have two gear tooth sensors that can be mounted to equalize the two sides of the chassis (Doug worked on this with interrupts two years ago and I still have the notes on flipping bits in the registers).

** We can also work on the input side – Jimmy and I have been chatting about tank drive (having two single axis inputs vs. one two axis input, which is really confusing). Karlin found that http://www.ifirobotics.com/ already has a USB board that makes it possible to connect all sorts of other input devices to the operator interface. The easiest way to do this right now is to take the two joysticks provided and to constrain their motion to only one axis (y-axis), read only that input and let a driver try using two of them at the same time. Fun problem.

** Goals for the week: The girls are planning to have the prototype chassis finished by sometime middle of this week. They are currently using wheels from two years ago and the same gear box and motors from two years ago. They had some issues finding bolts long enough today and are going to have to move their electronics board out of the way to the back in order to screw the gear boxes to the frame, but their close. The electronics look like they are all hooked up and can turn on (please note they are using the robot controller from 2005, which means when they compile, they need to make sure they compile for the PIC18F8520 (pre-2006) and not the PIC18F8722 (post-2006). They will likely also need help with the chain and tensioning it. They are also working with Ms. Mourad to get these omnidirectional wheels (http://www.andymark.biz/am-0083.html) Really hot stuff.

I’m really looking forward to debugging system issues on this prototype chassis (get all the code and sensors hooked up) before the real chassis gets built!! Go team.



LIFT / CLAW: (Nandini, Erin, Anne, Emily (Mentor), Karlin (Mentor)) Good progress this week. I worked with newcomers Annie, Divya and Crystal to build a portion of the overpass field element today (see photo below). This really helped us to experience what it like to manipulate the ball (like put it on and push it off) its sitting position. We discovered that it’s actually pretty easy to push off from below by just running into it with some momentum using first, Annie’s forehead, and then pieces of 80-20 and PVC pipe.

** In the last week, we’ve been bouncing a number of ideas. Jimmy and I have talked about elevator lifts and four bar linkages which help us start within the 60” tall X 38” width boundary and still be able to reach up beyond 78” with a claw. Fundamentally I am pretty concerned about the center of mass of the 40” diameter trackball. I think the more time we can drive around and keep it above our CG the better.

** Nandini spent a lot of time with David on Saturday exploring a good number of options for the claw. They explored the option of the claw having three actuated prongs, and then three prongs with only one active to clasp it. Nandini also had an interesting idea based on a tarp and a draw string, and finally, she has some really good thoughts on using rollers at the ends of the arm / gripper to keep the ball in place once in possession and to also expel the ball when ready. Thanks to Nandini for catching us up over the phone today and to Crystal for calling her.

** Annie, Crystal, Divya and I, after discovering the power of pushing from below, took a look at a solution based on a windmill. The pivot point would be high up and the two windmill arms tucked within the starting volume would rotate out to push and catch the trackball coming off the overpass all in one motion of driving through under the overpass at a fast speed. (VIDEO: Click here) . Divya came up with the idea of having a second smaller scoop type mechanism at the base of the robot that can get the trackball off the ground and into a position that the big windmill can scoop up and drop onto the overpass in order to hurdle. I’m having them prototype the system with ¾” PVC (they have a stash in the science project room in the back, along with a PVC cutter) now.

** So, I think all solutions are still on the table, but I would encourage everyone to start moving to wood, PVC etc. in 1:1 scale. I also pulled up Solidworks and used sketches to demonstrate the path of a linkage / arm and have them guide me while sketching changes. We also, with Doug’s speedy help, managed to get the portion of the overpass that we built at the right height (it’s tall) and sturdy enough for now. You’ll have to put it together outside in the walkway since with the ball on it, it’s way higher than the classroom ceilings.

OTHER STUFF: Keep us posted on your schedule. Thanks to Karlin for putting up the Google Calendar. I’ll be there on Wednesday night by 5:30 pm. See everyone next week for more fun!

Em

The Rack is Rising!

Thank you to all the Mentors and parents who have worked so hard over the past 3 days - even missing the NFL Conference Games - to build the Rack and the Crate. Doug, David, George, George, Richard, Beth, Debbie - you rock!

Parents Corner

The 2007 FIRST season is in full swing, and our Gatorbotics team is working hard. Parents, you can check here for the latest schedules etc. Just click on the links on your right to find out what each sub team is up to!