A plea for roboticists
 

I would like to start by saying that I am shooting for a Mechanical Engineering major, but I consider myself a roboticist at all times.

I officially stopped competing in FIRST the year before last, but I've been keeping up with the competitions. Drawing from the past 5 years of robot history that I've witnessed, I am of the opinion that Dave Lavery will come on stage this year, set down a big cardboard box, and say:

"This year, we've built the robot for you. Rule #1: you may not modify the kit robot. Now, for the next 6 weeks, would you please, please, PLEASE program it?!"

I now recall all of my previous years, and it seems like all I was doing was building big, glorified RC cars with arms poking out of their midsections. I worked in Olin's Intelligent Vehicles Laboratory over this summer as a motion control engineer (read: designed and built stuff that moved), and I honestly spent two months out of the three programming. We, in fact, programmed so much that we ran out of time to build the robot :yikes: - we're going to finish during the school year.

Meanwhile, in the robotics lab next door, three of my friends were working on a robotic biped named M2 that was built in 1998. The motion control hardware (motor controllers and amplifiers) has been through 3 revisions simply because the robot is so old! After 7 years of programming (4 of which were at MIT), we still can't get it to walk.

Ever since Autonomous mode debuted, FIRST has wanted us to make real robots, the likes of which are active in every university with a systems engineering major. Every year, however, I see a majority of robots either dead for 15 seconds or giving the double "254nownownow!" signal to both sides of their drive train. It took a year of college for me to realize how sad that is, given the massive amounts of tools the kit provides you (a fully tuned PID loop for the most powerful motors in the kit, an IR beacon, pressure transducers, current sensors...).

My faculty advisor at Olin used to be the vice president of engineering at iRobot, and he told me that iRobot's had around 40 job openings go unfilled for the past year or two, simply because of a lack of skilled roboticists. What is a roboticist, you say? A roboticist is the uber-engineer, the systems engineer on steroids. A roboticist can create mechanical designs as effectively as any MechE, that factor in those pesky things like wires, sensors, and effects like EMF. A roboticist is envisioning the control algorithm for their design while the design is still in Inventor or Solidworks. A roboticist selects the appropriate servomotor and control hardware for their design before any solid model is ever touched.

I'll stop rambling on now, and just submit this plea: please, be roboticists! I know it's really hard learning everything at once, which is why many teams default to just working on the mechanical structure of their robots. I promise you though - if you make the extra studying and programming effort early on, if you learn your sensors well, and if you take the time to research PID (proportional-integral-derivative) control loops, higher and lower brain functionalities, and even a little SLAM (simultaneous-localization-and-mapping), your robot will balance on its two rear wheels and thank you at the end of the season. More importantly, if you pick up enough along the way, you can go to any college you want, and any corresponding career.

Re: A plea for roboticists
 

Great point... there really is a profession out there for those of you who like to do a little bit of all of it!! And even if you cant get the dream job as a roboticist (is that now an official title you can job search for?!?! lol), you can become a systems engineer (what I am/was)!! You do a little bit of all of it... end up a jack of all trades (sometimes master of none), but it s NEVER boring!

Re: A plea for roboticists
 

This aspect of FIRST drove me nuts this year. I was one of two programmers on the team, and after we ditched the camera (with 3 days to ship), we immediately started work on autonomous mode coding. We finished at 10 PM the night before ship, but as anyone who was at VCU this year can probably remember, (I'm not trying to brag at this point) , we had the killer autonomous mode that capped the side goal every time in the qualifying and finals round except once. I also think we that we had one of the very few useful autonomous modes (at least at the beginning of the qualifying matches) That said, we were using dead reckoning, but it goes to show how much a (not even that well-coded) autonomous mode can go. I believe that upon watching a few of our rounds, several teams promptly copied our technique and adapted it for their robot, as a couple teams I was watching that originally had no useful autonomous mode suddenly were attempting to cap the side goals.

Just imagine if this year at kickoff, Dave's robot not only followed him and his shirt, but also capped the center goal on the demo field, not just in the animation. This would have been greatly inspirational to all teams (as all had to watch the kickoff), and might have resulted in the center goal being capped at least once this year. However, I can see why they didn't have the robot do anything except follow Dave: it would have been copied by teams so they could skimp on days and nights of intensive design. That being said, I did notice more than one team at both VCU and Atlanta that copied a design from the animation anyways.(forklift with a single bar in the center).

But please, I would really like to see at least something other than playing dead or the double-254gogogo from 80% of the teams. It's more interesting than watching both sides of the field park in place or ram each other head on.

And remember: a good autonomous mode is a good bargaining chip when trying to get picked for finals! :] Of course, if you have a good autonomous mode, it's quite likely that you've made it into the top 8 and are doing the picking. :p

Re: A plea for roboticists
 

As you've said, there's a large lack of programming going into all of our bots. I think this is primarily because it's just so hard to do anything useful in 6 weeks. You said you and a team of others spent two months programming a robot this summer. Full time. FIRST has 6 weeks of working after school. I personally am amazed that any teams out of the 1000 managed to do anything useful with the very finicky CMU Cam.

That said, there's obviously still room for some programming on the robots. I wanted to introduce my team to the wonders of feedback control, but I was sorely needed for just getting a robot built. As you also noted, it's very easy to spend lots of time programming. Having a well programmed robot that's not built is a bit of a problem, obviously. And many teams are hurting just for mechanical mentors and getting a robot built, period.

I suppose I'm sounding like a bit of a downer. I really like programming and digital controls and all, but I think there's only so much that can be done with a team of high school students and 6 weeks, short of mentors writing the the entire program themselves and just showing it to the kids. Expecting teams to design, build, and program a highly adaptable machine vision system in 6 weeks seems a bit much to me.

Re: A plea for roboticists
 

You have no idea how much this pains me!

I love programming so much, but in the two years I've done FIRST, I've never been able to do anything really interesting because I had to help with the more pressing need of building the robot.

This year however, things will change. ;)

Oh, and my CMU cam arrived dead!

Re: A plea for roboticists
 

As for coding for two months, this our labs' first year. Starting from scrach, we're putting 3 3.2 GHz processor, 2 GB RAM computers in one box, distributing computing between the three of them, talking to an ethernet based 8-axis coordinated motor controller, grabbing data from a firewire stereo vision camera, grabbing data from a RS232 GPS unit, and grabbing data from a RS422 laser ranging sensor. Just about all we had to write from scratch (transparent open and closed-loop motor control, data collection algorithms from our sensors, and lower-level communications protocols) has already been done for FIRST. It literally took 6 of us coding for two months to get to the point where the FIRST controller has been for the past 3 years.

But that's the thing - no FIRST team on Chief Delphi sticks to the 6 week time frame to build their robot. I don't mean to say that they build past the end of the ship date, but I instead refer you to the picture gallery. About every 4th picture is a new shifting transmission, crab drive module, or other personal MechEng project with FIRST resources. FIRST seems to have settled on a robot controller over the past few years, and even offered us an equally capable autonomous Robovation controller or two to practice code development - I don't think the 6 weeks reasoning quite holds up. I think if teams developed the same interest in good, solid programming skills that they currently have in creating (in my opinion) rather unnecessary gear-shifting transmissions which may or may not be useful in the upcoming year, we could have some really, really cool autonomous modes.

Re: A plea for roboticists
 

I agree totally. Of the people I interact with in the FIRST community, I only know one other person working on a navigation system.

Re: A plea for roboticists
 

Same situation here.

Re: A plea for roboticists
 

No offence people, but i think this conversation is the slightest bit slanted, as to how mostly programmers are on CD in the off-season :D. That being said, I've really wondered what it is like to be one of the programmers, I'm just glad I'm not one of them, as our team is very evil to them and gives them about 3- days to program the thing. Most of the programing done while we're building is "preparation code" that they run on past year's bot to test their programming with. And I'm pretty certain we're not the only team...

Eventually we hope to be like one of those teams that manage to build their bot in the first 3 weeks and spend the last half tweaking, but a lot is going to have to happen for us to get into that position.

On the note of programmers getting more time. Like any other aspect of the bot, all of the parts are vital to get it to work but when it comes down to it, teams need to decide which aspects get that bit of extra time and energy to improve. And this year, since the main challenge of the game was a pushing match, and a very difficult arm design, most teams spent a lot of time on that instead of on the autonomous code which many teams thought was just too much effort for the payoff (two more tetras? In finals the bots could do that in about 12 seconds

I think FIRST could encourage more autonomous programming, or time spent on creating it, by making it more central to the game. In '04 knocking the ball off was an incredibly useful auto code, that gave teams a killer advantage. But this season, we all know that wasn't quite the case :).

If FIRST would come out with a game that gave a difficult challenge in autonomous programming, one that would make a team considerably more powerful in a game (moreso than this year, it wasn't impossible, just very very difficult, for a very small payoff) a lot more teams would be giving more time to the programmers.

Anyway that's my $0.02

Re: A plea for roboticists
 

I personally think that FIRST simply does not give us enough time to program for all the sensors. Add to this the fact that we get a new sensor every year, and you get frantic programmers. 330 tried both IR sensors in 2004 and the camera this year, and we ran out of time (and weight) on both.

Let's look at various games, staring with 2003, when auto mode started. '03 had (I think) the largest proportion of teams with an auto mode. There were no IR sensors or cameras, nothing terribly complex. 2004 had the IR sensors introduced. I personally don't know of a single team that used them. 2005 had the CMU Cam. As far as I know, no one used it with any large degree of success.

I personally think that the only thing wrong with more sensors/more complex processors is that they need to be released early, perhaps at the same time as the game clue. Yes, let's have more autonomous robots, but can we have EIGHT weeks to get our sensor and autonomous programming done into the bargain? I think that if this is not done, few, if any, teams will have a sensor-driven auto mode in the future. More complex sensors+same time=more mars rocks because programmers have less time to do anything about moving the robot according to the feedback.

Oh, and some teams don't have very many programmers or can't spare students to learn programming because they are needed to build the robot. So they are already at a programming disadvantage, then you add something that even veterans have a hard time doing, and what do you get?

Re: A plea for roboticists
 

WPI offers first undergraduate "robotics engineering" major...

http://www.chiefdelphi.com/forums/sh...ad.php?t=50361

Re: A plea for roboticists
 

I would go so far as to extend this plea for roboticists to other technical universities as well. I know a lot of roboticists (by Gui's definition) at MIT, but they took it upon themselves to learn both sides of the coin. Mechanical Engineering and Electrical Engineering are two separate majors here and many students go through their four years of undergraduate education without getting much exposure to the other side. (Havng taken the famous 2.007 Mechanical Engineering course, formerly Prof. Flowers' 2.70, I can tell you that building those robots is a purely mechanical endeavor, with the exception, I guess, of soldering your motors to the connectors.) We have entire courses on control theory, but I've relied on non-coursework projects (like FIRST) to actually put the theory into practice.

I will echo Gui's argument: FIRST's control package is incredible. Spend some time with it in the offseason. Build a Segway, or an arm that can balance a cup of water 5' out while your robot goes over a speed bump. And teach your HS students how to do it; you don't need transfer functions and pole-zero plots to understand the idea of a virtual spring and damper. This will prepare them to take it on during the very short FIRST season. And, when they go to college, they will demand an education for roboticists.

Re: A plea for roboticists
 

Part of the reason also has to do with the fact that they dont hire people directly out of college. I'll have about 1.5 years of work experience but I don't think that would be enough to get the job right out of college. :(
The problem is that programming a robot does not involve just knowing how to program. Once you know how to program you have the problem of figuring out how to program a robot which is another annoying can of worms because there are many different ways to do so.

Re: A plea for roboticists
 

Reading this thread this year makes me realise again how far teams have come in 2006 for autonomous mode. Thinking of alliances scoring 50 points in autonomous would have been unheard of even a year ago.

Hopefully programming can continue to improve. I think FIRST made their point last year with it's importance.

Re: A plea for roboticists
 

How many points was hanging on the bar worth in 2004? At least one robot could do it autonomously.