17-01-2011 12:17

MattC9

Re: pic: DiscoBots Nano-killough

Nice!! it looks great! I'm just curious why y'all went with the omni side of this game?

17-01-2011 13:49

AllenGregoryIV

Re: pic: DiscoBots Nano-killough

We did a game analysis and came to two ideas we could either build a drive train that is more powerful than other robots or we could build a robot that is more maneuverable. We also decided that speed was a crucial component in getting as many tubes hung as possible.

With that we would either need to build a six wheel drive train with a shifting gearbox or go with omni-directional and have lots of speed and be able to get around defenders quickly. We also knew that when in the our zone we will be able to position ourselves to hang without defender interference unlike 2007. Strafing in your zone to be able to get to the proper peg without exiting your zone is important so defenders can not affect you.

Swerve drive may be a better option (to have more power and still get maneuverability) but we have never prototyped swerve modules so we went with what we considered to be the best method to accomplish our game strategy that we could complete in the six weeks.

17-01-2011 14:19

Stuart

Re: pic: DiscoBots Nano-killough

1745 did this type of design in 2008. it worked out well but I would suggest some sort of suspension. we found places in the carpet arnt as level as others and if all the wheels arnt applying the same pressure and traction to the floor it can mess with handling.

24-01-2011 15:56

J.Warsoff

Re: pic: DiscoBots Nano-killough

that is one interesting chassis. the only thing i'm worried about is stability. it seams that without suspension, as stated above, that the robot will be very prone to tipping over. Use suspension and distribute the weight effectively, and you will have one heck of a robot there!

24-01-2011 16:42

Bochek

Re: pic: DiscoBots Nano-killough

Having built a robot like this last year, i woud warn again the omni drive configuration... we realized (too late in the build season) that even the smallest bump or unevenness of the playing field would cause the robot to turn violantly, no matter how much the software tries to compensate for it.

- Bochek

24-01-2011 16:45

Chris is me

Re: pic: DiscoBots Nano-killough

Why would a robot without suspension be more likely to tip?

24-01-2011 17:02

AllenGregoryIV

Re: pic: DiscoBots Nano-killough

I am not so sure what the tipping thing is about either, our chassis is 1.125" from the ground and we will have bumpers extending from that. Tipping over the bumpers will be pretty difficult.

However the traction problem is something we are definitely considering. We have several solutions in mind both mechanical and software to try to fix any problems that we might have.

Thanks for the comments and tips.

28-01-2011 14:32

J.Warsoff

Re: pic: DiscoBots Nano-killough

It won't absorb the force of a hit from another bot as well. Plus, suspension will keep it stable while maneuvering at faster speeds.

Also guys, a rookie on my team was wondering how stable this will be when you mount the other parts like the elevator, minibot/minibot deployment, etc. on it?

28-01-2011 16:38

AllenGregoryIV

Re: pic: DiscoBots Nano-killough

Stability is something we are definitely testing. One of the things we know is that we should never be driving at high speeds with our lift up. So the vast majority of our weight besides the outer lift rails will be low to the ground when driving.

Stability is a problem every team has and our frame is the same length as most teams narror dimension.

28-01-2011 17:41

J.Warsoff

Re: pic: DiscoBots Nano-killough

Sounds good. Here's another tip that helped my team last year: make your center of gravity very low, so put all your heavy stuff towards the bottom of the robot.

30-10-2013 18:54

gpetilli

Re: pic: DiscoBots Nano-killough

How did this work out for you? We are building something very similar as our off season project. This style chassis is extremely stiff and has a very low center of gravity so I doubt you had the predicted stability problems nor needed a suspension.

The major difference with ours is we went asymmetric - meaning we put the wheels at a 30deg toe-in instead of standard 60deg. This gives us more usable torque front to back (87% vs swerve) and less lateral (50% vs swerver). 45deg would give 70% vs swerve in all orthogonal directions.

Did you do robot centric or field centric controls? What sensors did you use and would you use them again?

30-10-2013 19:18

AllenGregoryIV

Re: pic: DiscoBots Nano-killough

Haven't seen this thread in awhile. I was a discobots mentor when we built this robot and was in charge of the control systems team.

Was it a fun challenge, yes. Would I ever build it for competition again, NO.

You absolutely need some form of suspension or you will be struggling to get all your wheels to touch the ground at the same time. If you use driver centric instead of robot centric controls it helps alleviate some of the problems with wheels slipping. Even the smallest imperfection in the floor will make your robot turn drastically, the plates around the mini-bot poles in 2011, were a huge problem for this drivetrain. The only sensor we really used was a gyro, to keep us oriented. We had "Halo" style controls, so one joystick was for strafing and forwards and backwards and the other was for rotation. We also implemented quick position buttons on the top the turn joystick so you could quickly go to any 90 deg offset from strait ahead.

By our 2nd event, we had it working well enough for us to be the 2nd overall pick and make it the finals before losing to 118 and 1477.

Very few games allow for you to give up that much of a traction advantage and still play well.

31-10-2013 21:21

Jeffrafa

Re: pic: DiscoBots Nano-killough

Although its an old thread, I thought I would chime in with a little info and our teams experience with a similar kiwi drive this past year.

A 4-wheel Holonomic, as 2587 did in 2011 has similar issues encountered by mecanum when it comes to maintaining contact and similar weight distribution between all 4 wheels. Without equal contact throughout a movement, motion will become unpredictable - this can be mitigated with a decent closed-loop control system using a gyro sensor or other feedback, but you are just trying to overcome an inherent mechanical issue.

The 3 basic factors are the floor flatness, frame stiffness, and (optionally) the use of suspension. obviously 4-wheel mecanum drivetrains are much more common than 4-wheel holonomic, so it is easier to look there for successful platforms. Essentially, most successful mecanum bases have either a flexible frame (like a kitbot C-Base), or they have a stiff frame and some sort of suspension. On very flat fields without any 1/4" lexan to drive on some teams may have mild success with a stiff frame and no suspension, but this is not ideal.

An interesting early 4-wheel holonomic used in FIRST competition is 116's 2005 robot: http://team116.org/our-team/robots/2005-robot/ They did a few unique things, including 82-degree cambered omni-wheels (this accomplished 2 primary things - pushed their contact point right to the edge of the frame for stability, and put more rollers in contact with the ground at a time for a much smoother ride on the older-style AM trick wheels). With dual roller omni's available today the second benefit isn't important, but the 1st one is still pretty cool. They had a pretty stiff welded frame, and I can't tell if they had suspension - but it was a very flat field and I think they at least had an adjustment to make sure all 4 wheels were planar.

Last year we exercised the flexible frame perimiter rules to our advantage, chose to do an omni-directional drive, but side-stepped many of these potential issues by going with a 3-wheel holonomic (sometimes called a kiwi drive). Because you only have 3 points of contact, they are all always touching the ground (remember, 3 points make a plane, and a 3-legged stool will never teter). It was the first time we had built something like that (had experimented with mecanum before), but it was a huge success for us. We attended 2 regionals and championships, ranked 4th in our division and made it to semi-finals (our best CMP showing in our 10-year history). In 44 matches throughout the season, plus another ~25 in off-season events we never once changed wheels, or performed any transmission maintenance - the drive was completely trouble-free.

We definitely didn't play defense much, but with a skilled driver we were pretty good at quickly maneuvering around the field and outrunning/avoiding defense. We used robot-relative control (not field-relative, but our driver had RC experience and a fair amount of drive practice). We only used one gyro, which greatly improved our rotational control, and was pretty simple to implement. If the frame perimeter remains flexible, we have a flat field, and we value maneuverability over pushing power I would say we might likely re-use this drivetrain, but in other situations it might not be ideal (fitting it in 28"x38" may be a pretty big stability sacrifice).

31-10-2013 21:48

DampRobot

Re: pic: DiscoBots Nano-killough

Wow... that's a very cool design. I presume it was for a 30pt climber. How did it work out?

02-11-2013 13:19

gpetilli

Re: pic: DiscoBots Nano-killough

What was the magic change that made the second event better? 2nd pick is fairly good, obviously your arm must have been great and your scoring was probably enhanced by the strafe capability.

03-11-2013 11:12

AllenGregoryIV

Re: pic: DiscoBots Nano-killough

Getting the drive code under control, autonomous was about 100% (we used ultrasonic sensors to get to the pegs), arm and claw control improved, minibot was drastically improved, and more driver practice.

We also increased the speed of the drivetrain by changing the gears in the tough boxes, that helped a lot.

10-11-2013 17:57

gpetilli

Re: pic: DiscoBots Nano-killough

do you recall the gears before and after? I think earlier in the post mentioned 10.7:1. We have been using 12.7:1 with 6" wheels and that seems plenty fast. Were you going for higher top speed, more acceleration or more control?

11-11-2013 02:20

AllenGregoryIV

Re: pic: DiscoBots Nano-killough

I believe we started at 10.7:1 and moved to 8.45:1. In 2011 we ended up having to go across half to 3/4 of the field regularly to get game pieces and the extra speed helped a lot.

At 12.7:1, unless my math is wrong, your actual speed is going to be under 10 feet per second. That's not fast by FRC standards. It really depends on the game and what you want your robot to do.

11-11-2013 09:58

Ether

Re: pic: DiscoBots Nano-killough

11 feet/sec at motor free speed. So yeah, actual top speed probably well under 10 feet/sec.

11-11-2013 11:33

gpetilli

Re: pic: DiscoBots Nano-killough

We are using the 12.7:1 gearboxes mainly because we did not want to spend money on an off season experiment. We need to better understand the limitations of Killough drive.

Two things we are doing that modify the standard calculations.

1) We plan to use MiniCIMs which have a free speed of 6200RPM (16.7% faster than CIM). Given that we don't intend to get into pushing matches, closer match to Jaguar current capabilities and the reduced weight, this seems like a trade we thought we should try.

2) According to Ether's kinematic calculations, by toe-in the wheels we lose cos(30) in torque but gain 1/cos(30) in top speed. This is another 15%.

At face value 10fps * 1.16 * 1.15 = ~13fps. Our crude estimator spreadsheet predicts 12.5fps front/back and >18fps lateral (1/cos(60)). We are planning to instrument and measure the step response on Thursday (students tell me the code and electrical are finally ready). I will report what we achieved by end of week.

Assuming we hit >12fps, do you think that would be competitive? If we are able to control it, and the game warrants Holonomic we can always change the gears in the TBnano to a lower ratio in the production robot.