The students at Exothermics been having some spirited discussions about the best drive system for the Programming Skills Challenge (PSC). One of the key issues is the drive system and the question of whether or not it is possible to build a line-following holonomic* drive. Omni drives (swerve, holonomic, or side-strafing like 148 last year) would make it easier to dump over the wall when being defended, but would also make autonomous more challenging. At least, that’s the argument.
So, over to you. Are holo drives a reasonable approach for the PSC?
Line following with three sensors says, if the left and center see the line, turn left, when the right and center see the line turn right. This presumes a "front", and a holo drive has a front, so it should work fine.
There are details, do you turn the robot to follow the line or just “crab” to follow. Since the holobot can turn in place, following a line that makes a sharp turn can be done by rotating the robot.
An interesting point is how many line sensors do you use? Three or five? The use of 5 sensors would allow any “face” to become the “front”.
So I would say, yes to the holobot. (But then I’ve always liked them)
I would think programming a holonomic to do line following could be easier than normal, since as the line tracking starts to get off, you simply strafe left or right instead of turning.
This way you are always going forward and not turning.
Strafing sounds like a really good idea but couldn’t it get out of hand if your robot has an uncorrected bias to turn left or right, as they all do. Worst case I could see a robot arriving at its destination, on the line, but sideways (well almost!)
Team 508 (placed 3rd in the world @ Elevation) used a holonomic drive (my design) and was able to somewhat successfully track a line. Although, due to the three days we had to program autonomous, we did not end up using the code. With some trial and error and more time what we had attempted could have been refined. Our system used the Vex accelerometer with a combination of three light sensors. One cool thing about a holonomic drive is that getting the robot back on the line is easier.
The hard(er) part is that the drive works on the basis of slippage. This means that going strait doesn’t mean that the drive goes exactly strait, there is some “noise” in the direction. This “noise” must be compensated for or you may miss your target. With clever programming, sensors and checks it can be done but expect to spend alot of time doing this.
My opinion, a holonomic drive is great for competition but for a one minute autonomous skill challenge, as the robot progresses this noise wil amplify and cause the robot to loose it’s track overtime. Dealing with this challenge will take skilled programming and lots of time. if you dont have the time/resources to do this then simply don’t. If your going for gold and put the right effort into it there are complex ways to make it work.
The concept of a holonomic drive turned my brain upsidedown. I love omni’s and have built several holonomic drives. One day I hope to build a holonomic car and a holonomic Battlebot. Point being, I am dabbling with an idea of mine using ultrasonic “radar” which could prove to be deadly
The simple concept of using radar with such a powerful drivetrain is just… evil but awesome!
Wow, we have a holonomic chassi which we want to use for a comp bot but are unable to because our programmer is making a “radar” on it. He has done it by attaching the ultrasound sensor on top of the large 64 tooth gear which is powered by a motor underneath. But he hasn’t really had enough time to fully program it. It works, kinda. But because now we’ve started building we’re expecting to program for us non stop, so his project has been canceled and most likely be taken apart soon.
And about using holo in auto, the new college challange rules. 60 sec auto with 80 sec driver control. And the only college teams that wont be using a holo will be the teams that cant fit them on due to the rest of the robot, so line tracking with holo will be more important then you think.
Why put the ultrasonic on a motorized gear when a holonomic can pivot on it’s center? hehe…
Also why only one ultrasonic? More resolution is always better and the refresh rate of the radar is critical.
I am both humbled and angry at this, your probably right but I liked being one of the few teams using a holonomic drive! I guess those days are over. And because I put a holonomic in the animation, everyone is going to see it and attempt to build one… curse myself! Oh well
Let me set the record strait, just like before - It can be done, it is just difficult. If any teams runs on the schedule my old team did (about three weeks allowed to build a robot from design to comp. day) then you will have trouble).
An ultrasonic does seem interesting but can you rely on its accuracy for location programming? i don’t think the refresh rate is reliable for a moving robot. it is a possible endeavor though.
Last year at Pan Pacific we tried to use an ultrasonic sensor in our autonomous. However we needed a delay of 5ms to get an accurate reading. On the holonomic front, it took me no more than a day to design and build. There are many benefits to this drive system; however, there are many logical drawbacks. Such as you can only effectively employ a 4 or 8 motor drive. Unfortuneatly with four motors, there is a greater tendency of a drive base to fail. 508 had first hand experience with this. We had a great run in dallas until, amongst other things, one side of their drivebase went out in finals :(. When that happens the only way you can efficiently go is in a circle. So make sure you know what type of load you will be putting on these motors, and don’t forget a holonomic drive base is drastically less efficient and should have a high factor of safety.
Our programmer said the programming was quite complicated so not many teams would use it, only a minute after he said that I saw that the robot on the table behind us was using the same chassi as us… Although, couldn’t they just get the programming of google, or somewhere on this site… But you would be suprised how many teams don’t use it but have the space to.
Our team ran a one week schedule last year, started with nothing a week before nationals. Ended up coming second with that robot and another of our robots.
We tried using ultrasonics in Elevation. Unfortunately, it does not work very well. We had huge troubles with the sensors and ultimately ditched them for shaft encoders (which actually work very well for autonomous). Trying to make a radar is going to be practically impossible.
Perhaps I’m not understanding this right, but I don’t think you’ll have much luck with that (the signals will just interfere with each other). Also remember that if the signal bounces off a non-perpendicular surface (think the triangular goals last year), you won’t get a return at all.
I’ve never tried this, but simply applying standard tank line-tracking code to a holonomic/kicker drive doesn’t work? The kicker drive should work at least, if you ignore the kicker wheel - it will just have to compensate for the lack of normal (non-omni) wheels.
What non omni wheels? Both pictures use all omni’s. But we’re probably going to make slide drive soon, and since we want even better autos this year line tracking will be essential.
I was referring to the fact that omni drives make less predictable turns than drives with two non-omni wheels, because you can’t exactly predict their center of turning (which makes programming harder).
Ended up coming second with that robot and another of our robots.