Strap in kids, lots of information in this next post here, so I’ll try to sort it as best as possible:
There are 4 major types of intakes that I know of, but I will only be discussing two, the two wheeled ones (though the third is definitely the most modular and can be used most indiscriminately have no experience with it so hopefully someone else will be able to explain the best practices.) First though, some general best practices:
Speed: You want the linear velocity of your intake wheels to be at least double your drivetrains maximum speed. This allows you to hold on to objects when they are driving in the opposite direction and allows you to pick up game pieces when moving.
Compression: As stated above by Andrew, something in your system has to give. The stiffer a gamepiece is, the more compression is required by your robot. Objects with no give, e.g. totes, will require a lot of compression, objects with slight give, e.g. frisbees, require less but still moderate compression, and objects with lots of give, e.g. boulders require very little compression (and often times components can be hard mounted.) The most common ways to achieve compression are through wheel choice, often teams using andymark compression wheels or Fairlane wheels (especially their “suregrip” wheels), and through spring loading/surgical tubing components. Some combination of these two may be required for some tougher game pieces.
Wheel type: This varies greatly dependent upon the intake, gamepiece, etc. The best advice here is to pick up a couple of every wheel you want to try and just test them all.
If you aren’t willing to dish out for the prototype, you shouldn’t be willing to risk dishing out the money for the final. (Also nothing to stop you from using these wheels on the final robot.) General list of wheels to test: McMaster Blue urethane, McMaster Suregrip (any type, though I have found yellow urethane works well), Andymark Compression, Banebot wheels (all three colors).
- Vertical
When I was a student on 1086, this style of intake was by far our best. Whether they were our alliesor opponents, we were able to steal gamepieces from the best teams in the world. What made our intake as good as it was mainly these concepts:
Points of contact- First of all, you want your leading edge to be the wheel. This seems fairly obvious, but you want to give the wheel a slight amount of room to get a grip on the gamepiece. Secondarily comes the angle needed to capture the ball. As you can see in the diagram below, through our testing we found that at a 30-45-degree angle with the level line was where we determined the best point of contact existed. Ideally in a perfect world you would try to make your intake such that when your intake touches the ball, the 45-degree angle of your wheel is colinear with the 45-degree angle of the ball (see diagram for better understanding). Unfortunately, because you need compression in either the ball or the intake, this method (while it will almost always work) is not the most efficient nor most effective method. This is where prototyping comes in. What you will want to do is start your intake at the 45-45 height, and slowly work your height down until you find the height that suits your intake best. You want the lowest angle possible to get as much compression as you can to “own” gamepieces.
Sturdiness- For this intake you want it to be as durable as possible. This can vary greatly by whether you want a very flexible intake e.g. 1678 in 2014 or something very stiff like what we had in 2016. It’s obvious why you want a sturdy intake, but these intakes often take harder hits than horizontal intakes as they have less/no give in the horizontal direction.
Width - For this type of intake you want as wide an intake as possible. The more space you have to pick up, the easier it is to pick up. On our 2014 bot, we often joked about how we had the most intake space in all of first. While those intakes had a couple, issues preventing them from being as good as we would like, no bot I’ve seen was able to compete with us in this field.
This gave us as much space as possible to grab the ball and gave us a large advantage when trying to efficiently pick it up, especially on the large open field.
Best Examples:
2014: 2481, 1678, 254, 1114, (and 1086 :P)
2016: 148, 254, 118, 971 (and 1086 :P)
- Horizontal
This type of intake is a bit more finnicky than a vertical intake but can be very effective for both ball shaped and “oddly” shaped gamepieces. Unlike vertical intakes, which generally rely on friction with the floor to roll gamepieces and compress gamepieces, horizontal intakes do not require any outside force to properly intake. They also tend to be easier to realign gamepieces as they are being manipulated. While our 2015 intake had a lot of room to improve, here are my main takeaways from that year:
Points of contact - Unlike a vertical intake, which only has one major point of contact, horizontal intakes have two or more points of contact. The angle at which the wheel touches the object is significantly less important than the vertical intake. Instead, compression is very important based upon these points of contact. The first points of contact should be the two points which have the most area to move. As the gamepiece continues down the intake, you want progressively less give in your intake to funnel the gamepiece and grasp it more firmly. This makes it so that you will still have the speed of any other type of intake, while being able to own the gamepiece. A good idea I’ve seen a lot is adding a piston/servo to tighten the innermost contact point so that you optimize speed when intaking but can get enough grip to properly hold the piece with no risk of dropping it.
Wheel count - Typically what I’ve found is the more wheels the better. More surface area means more on the gamepiece, which means more friction resisting letting go of it. More wheels also generally mean it’s less likely for an object to get caught and not turn into position properly. More wheels = more better.
Wheel choice - Banebot wheels seem to generally be the wheel of choice for outer “fingers”, as they seem to be great at getting a solid grip on objects. So long as the “fingers” have a sufficient amount of give, the eraser-like rubber used in Banebot wheels seems to be great at providing good traction with the gamepiece. The inner wheels that we’ve found the best are suregrip wheels, but as with most things stated here proper prototyping is required to determine what you want (this goes for finger wheels as well.)
Best Examples:
2015: 1114, 254, 118, 2056 (look for good landfill bots basically)
2016: 401
2018: 1323 & Co. 148, 5406
- Gripper/Claw
I personally was not involved with our gripper in 2011, so I can’t give too much insight into what you should have to make a good gripper. Hopefully @electronica1 can give some insight, or someone from 330/217/25.
Best Examples: 330 2007/2011, 217 2011, 25 several years
- Vacuum
Again, haven’t prototyped/used one of these so hopefully another team (558/1102 maybe?) can give some insight into some best practices for this.
Best Examples: Haven’t seen any particularly effective ones.
Hope that this helps, anyone with questions/comments feel free. 
