What should we get? The grant budget is about 2,000.
Scouters have asked if we use gyros of any kind but we never have in the past 2 years. What is it? Benefits?
Also, listing any ideas would help. We’ve added the limelight vision camera to the list. I did hear about a 2.0 version but when will that be?
A gyro has a lot of benefits. You can use it to turn to an angle with some sort of closed-loop control like PID, instead of dead reckoning everything with just encoders. Furthermore, we’ve been using our gyro for our robot’s odometry to determine our field-relative position (x, y, heading), which we’ve been using for path tracking in autonomous.
Theoretically, you could get your heading with encoders, but the math for that is complicated and it’s far more complicated than just buying a few good gyros. We’ve had success with using the navX-micro, and we haven’t seen the need to buy the full navX – it’s the same chip, so the performance shouldn’t differ that much. I’ve also heard good things about the Pidgeon IMU from CTRE, which is cheaper than both the navX and navX-micro, so definitely look into that.
We’ve personally had pretty bad experiences with the ADXRS450 gyro since we’ve found that it drifts quite a bit over the course of a 2:30 match. Your mileage may vary.
With respect to everything else that you may be interested in purchasing, it’s never a bad time to stock up on VersaFrame + gussets and spare VersaPlanetary parts. These parts often go out of stock in-season, so it’s a good idea to have some stock in the shop.
This thread from earlier this year was full of neat suggestions that would fit your budget.
A trap that a lot of people fall into is to say “X is cool and useful therefore our robot should have one”. This is the wrong way to design things. The right way is to have a problem that you’re trying to solve and then figure out how to solve that effectively. Doing things the other way is likely to result in both lots of wasted time and reduced robot performance.
Be warned that cameras are the most common part for teams to put lots of effort into without useful results.
As for what you should get, what are the limits of your grant? Does it have to be robot parts? If not, the suggestion to buy tools is a good one. You might also consider using part of your budget to buy some very boring-sounding things like aluminum extrusions and bolts just to have them on hand. Do you have space to store big things?
A camera with high fps is necessary for us. Our only other camera just lags too bad to even be used as one.
Im not too sure on the tools part but im sure my mentor would put them down if needed. We have many tools of the same kind.
If the grant terms will allow it, I’d say wait until closer to the holidays. AndyMark, REV, and VEXpro all tend to have big product drops around then, sometimes supplemented with additional items after Kickoff if there was something that needed to stay under wraps.
There are always some risks to buying brand new products on a robot, but I think it’s best weighed item by item once you see what’s possible. (For example, I am naturally leery of most new gearboxes, but the 57 Sport was mostly an iteration of the BaneBots P80 so that made the risk more acceptable in my book.)
Are you sure this is a fps issue, and not something else? What type of camera is it?
^^ This. If the problem is fitting the signal in a network (or other) bandwidth, a higher fps or resolution camera is more likely to increase the lag than reduce it.
Just to pile on here a little bit, what have you been attempting to do with your current camera? There are at a lot of different things that teams do with cameras:
It sounds like you’ve tried to do 2, and would like to do 1, but that doing 2 has given you super-choppy results. Is that right?
How high do you consider a high frame rate? There are some computer vision tasks for which super high frame rates are beneficial but typical FRC autonomous tasks are not among them. Most teams will only find frame rates higher than an average webcam produces useful for 5.
I can try to get a picture of the camera. Not exactly sure what the fps is but if our bot is driving straight at constant speed, the feedback image shown on the driver station will be jumping frames about every 2 feet making it unrealiable to use. This is one reason we didnt have a cam for 2018. Weight was the other. This camera had long wires with it. Will update with a picture asap.
Limelight is so easy to use and robust that I don’t see a compelling reason not to buy one, but it depends on what you have already and what you actually need. LL can be used as a driver camera and a simple vision system, so it also saves you from wasting time with custom vision applications.
Versaplanetaries are universally useful, and run out of stock in-season every year. VP Encoder slices are also very useful for any year.
Motor controllers are nice.
Extra aluminum stock is always useful.
A set of rivets from blindrivetsupply.com is also handy, if you use rivets. They’re very cheap compared to other sources.
Extra tooling and tool organizers is always a very useful place to spend money. If you find yourself looking for screws all the time, invest a few hundred into organizers that will last a few years.
One of the things we found was that for a stream to support driving and close-in tasks (low goals, hanging gearx, getting through defenses, taking in game pieces), cutting resolution to increase frame rate and decrease latency was usually a good trade off.
Back to the original question: concur about upgrading tools; they will typically improve a team over multiple years. Metal stock and other generic hardware would be good. Having optional gears for your existing drive gearboxes on hand (maybe not every speed, but enough spread out for prototyping) is also helpful.
To be clear, the primary use of your camera is simply streaming video back to your driver station, correct?
Limelight is not the proper investment for that purpose. Limelight’s real strength is vision processing. If you have a robust controls/programming team that can handle those features, it may be worth getting. But if all you desire is streaming video to your driver station, you can achieve that for a fraction of the cost with a cheap USB webcam.
The issues you’re describing regarding your camera are not likely the result of a low FPS camera. Rather, there’s something else in your controls system that’s causing camera lag.
Vision processing… Like the way 195 used a camera for '17 SteamWorks? They programmed their camera to detect the reflective strip of tape on the high boiler for aiming when they shoot fuel up there. What i learned was they had a free moving turret to automatically adjust to the boiler. https://www.youtube.com/watch?time_continue=498&v=B9oDEedmoGc just keep an eye out for the turret on 195.
That would be correct. Also: 2016 tower openings, 2017 gears, 2018 power cubes on occasion… pretty much if you can see it in a bright or reflective color you have a chance to track it.
The camera is microsoft lifecam hd 3000 on andymark. Says it’s 30 fps.
It sure can be, but not at maximum resolution as others in the thread have mentioned. What I haven’t seen mentioned is the resolution you’re actually using; sharing that may help people determine whether it’s a simple configuration issue or something more interesting.
I don’t know the resolution
http://i.imgur.com/hBGnCkd.png
https://i.imgur.com/EttkLQ4.jpg
Ditto on this. Get the ADXRS453. Registers are the same as the 450 and as far as I can tell all the 450 code works fine with it. The 453 just adds “Ultrahigh vibration rejection: 0.01°/sec/g” which has made a world of difference in our experience.
If you’re not going for a pre-canned solution like Limelight, remember your camera & processing first needs to have low latency (aka lag). High framerates can sometimes help you get to this. But, a low-latency 10FPS is generally easier to work with than 30FPS delayed by half a second.
This is starting to diverge from the OP question, feel free to split off to another thread if desired.