Just curious, in competition, which one is used more often, autonomous drive or manual drive?

Just curious, in FRC competition, which one is used more often, autonomous drive or manual drive? Which mode is better?

Thanks!

Do you mean during sandstorm or in general?

Statistically, autonomous is used for 15/150 seconds and teleoperated is used for 135/150 seconds a match in most recent FRC games. In reality though, automating as much as you can to control the robot and execute strategies better, is a good idea. The “ideal mode” of control for most teams is that your drive team controls what the system autonomously attempts it on it’s own.

This enables insane things like auto-scoring for hatches, almost all good 2018 robots elevators, fast-lineup for gears and fast shooters in 2017, any-location high goal shooting in 2016, essentially every robot in 2015, and further back yet.

In short; if you can’t automate robot functions to be done on their own, you’re at a competitive disadvantage.

If you’re asking about sandstorm specifically for 2019, it comes down to your teams level of execution and capability.

1 Like

I mean in general, not only during sandstorm. Because some good teams control the robot so precise and quick, I doubt can manual drivers do such good job? Especially when the robot goes to the other side of the rocket, it does the work precise and quick.

High caliber teams are typically able to achieve a high degree of autonomy, automatically placing game pieces when near a target, such as the rocket this year or the airship in 2017. However, the vast majority of teams don’t have the resources or experience to successfully integrate this sort of tech in a way that’s superior to simply aligning with the target manually.

1 Like

That’s a good example of where autonomous functionality can seriously better your competitive ability. Teams use control theory to tell their elevator to go to a height, and it goes there and they sequentially go “okay at height, go forward slowly given vision to score a hatch”, or just “stay at height”.

Unfortunately Solomon is absolutely right, a lot of teams don’t have the resources (time, money, effort, knowledge) to execute it well. Good programming and systems integration is the difference between a good team and a great team.

To answer your question blatantly, yes it is a human controlling the motion of the robot. It might have some automated movements and features for the elevator, arm or other mechanisms, but in the sense of controlling the core driving motion of the robot throughout the match, yes, it comes down to drivers and practice.

1 Like

So people get better at controlling robots with practice, which is why it’s important to have your driver practice before the season. That may account for some bit of this.

There is also some things you can do with the controls to make the robot easier to drive. Deadband and nonlinear response come to mind.

After this, each bit of autonomy you add can see returns. Teams add position PID on their elevator so that they don’t have to manually control it. They might switch their drive train to close loop control so it drives smooth. They add a vision system so that they can line up game pieces. They may even add autonomous paths they can use when there is no defense.

I used a lot of terminology here so feel free to ask for clarification.

We’ve been improving as the season has gone on. At our first regional, we would get a hatch in auto and then manual drive the rest. Second regional, we had a better sandstorm auto and had some driver assistance with cameras. Heading into Houston, we’ve got a ton of stuff, and technically, if we aren’t defended against or bump by team mates, our driver could never touch the joystick and we can do two hatches in auto and then fill a rocket.

Of course we are expecting defense, so realistically we assume 25% fully-auto and 75% manual driving with camera alignment assist.

Do you mean that filling the rocket is also autonomous? I haven’t really seen anyone tall about this yet but it seems viable given the consistent scoring locations and vision targets at both ends.

1 Like

Yes. Our Operator has a touch screen monitor, and it has the field on it. He touches a scoring location (ie: right, top, far, rocket) to go for. If our robot is lined up at the feeder station, it grabs a hatch or cargo, then does the entire path and scoring, and goes back to the same spot at the feeder station, ready to go to another scoring location.

So if we don’t get bumped too badly, we can chain them together and it’s way faster than manually driving. However, we expect to be bumped a lot, so we intend to mostly use that feature where we’re paired with a more intimidating robot than ourselves, so we’re not the target of defense and can do our thing.

7 Likes

That is really freaking cool. I’ll definitely be watching you guys in Houston. Hopefully you can get the fully auto rocket to happen!

1 Like

We don’t use any pure autonomous during sandstorm or teleop. However, we (the drivers) have no manual controls over any mechanisms. Our elevator and end effector go to preset positions based on what level the operator selects (1, 2, 3, or cargo ship) and our hatch effector can automatically grab hatches and release them when we push against the rocket or cargo ship. And we have vision targeting which is really nice now. The most manual control we have is the drive train (swerve) and our climb, in which the arm that

Anyhow, this makes driving really easy once you know the controls. The only downsides really are that the software is much more complicated than last year so our first 2 competitions were really buggy. Also with all the functions for the operator I don’t have any more buttons on my XBox controller so software has given me a ‘shift’ key so that we can keep packing in more functions…

1 Like

This year during sandstorm, I think I’ve only seen one team in the PCH district use autonomous instead of teleop.

1 Like

While a lot of “make it work fast” stuff is implemented in autonomous or semi-autonomous code, there is also a lot which can be implemented in hardware. One example I read about this year involved polycarbonate “trapezoids” which engaged the keyhole slots above and below each hatch and HP loading station and only required the driver to align within a couple of inches, but then driving the robot forward automatically achieved alignment to within a something on the order of a sixteenth.

I have also seen teams (mostly in other games, though it should have worked this year and I would be surprised if no one did it) who implemented simple but smart telemetry on the robot which enabled the driver to do things which look like magic. Sometimes, it is as low tech as drawing lines on the driver station screen which the driver uses to align to the scoring or pickup stations; this year, it would likely have been two rectangles oriented to align with the retroreflective tape.

At the “even simpler” level, teams have figured out where their robot launches game pieces, then aimed a high-intensity, narrow-beamwidth flashlight at that spot, and set up shots by putting the spotlight in the goal (as viewed from the alliance station) and shooting.

There’s more than one way to skin a cat; do what works best for you.

3 Likes

I relate to this on a spiritual level, when we had the hatch pickup on us I had 23 distinct actions with left trigger as the “shift” jammed onto my controller.

Thanks for all replies! I learned a lot!

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.