Sensors Not Required: FRC Design "Sans Feedback"

[DjScribbles]

I think there is already enough incentive to use sensors.
This year we didn't have time to setup our shooter speed control on the robot until half way through our first district; the change in firing rate alone was jaw dropping (going from a 1 sec delay between shots to allow the wheel to spin up, to a .4 sec delay), add in the additional reliability (since an open loop would overshoot if left running too long, or undershoot if fired too quickly) and you can make a really strong case that sensors are an important part of the game.

Before I joined our team as a programming mentor, the team was heavily mechanically oriented, and they didn't use many sensors. Last year, I took my first swing with an encoder on our roller wheels (short range vertical shots), but not until we reached Michigan State Champs; this year we built a US1881 Latching Hall Effect sensor and a couple ring magnets for speed sensing, along with a gyro to keep us driving straight. We've also toyed around with PID controls and such, but haven't gotten enough expertise to use them effectively.

I think it is good that the games remain accessible to teams that don't have strong programming and controls abilities; however building your capabilities (in-season or off-season) has some pretty huge benefits in competition, but also in learning opportunities.

We're very lucky to have Mr. Ether nearby to bail us out when I get in over my head, and to dangle carrots out there for me to chase

[Zuelu562]

3623 used photosensors last year in a futile attempt at an automated pickup and shooter system (in retrospect we designed the mechanical system poorly. I joked "we forgot about inertia"). This year we used two limit switches, and prevented the system from overshooting certain points.

I've always wanted to get automated systems as a programmer myself, but the reality is I'm working on making sure the system as a WHOLE works. It's definitely viable to not use sensors and have no automated systems. I've had robots reach the semi-finals with less sensors than people on the drive team.

Automated systems are cool, and definitely can bring in kids, but don't overreach yourself just because you think you need to automate systems and have more sensors than the average automobile. If you want to put an automated system together, the offseason is a perfect time to try it out (assuming you can get kids and resources).

[GilaMonsterAlex]

3785 doesn't use to many sensors. This year we tried to implement a gyro to sense the angle of our shooter, and the camera for distance to run some simple calculations to be able to shoot for the 3pt goal easily and reliably but the processing caused our robot to lag. So we went the mechanical route, and had great success.

Since it's our 3rd year, I think we're progressing naturally in our abilities due to experience. But at the end of the day we need a robot that works, and we'd rather have a robot that can play the game well with manual driver controls, than one that marginally works but does so semi-automatically.

I will say that this year we grouped operations together so that one button would perform two tasks. We used timing to perform the operations correctly. I think as long as you keep trying to push the envelope, you're going in the correct direction. Whether you implement the actions or not are a team decision.

[stuart2054]

Let strategy, your goals and the mechanical design guide your use of sensors. Stay true to the KISS principal. In 2011 we used the AB photo eyes for autonomous to follow the tape and an ultrasonic prosimity sensor to give us distance to the wall to know when to hang the tube. Last year we used vision to set the shooter speed and indicate the aim was centered by turning on a light ring on the back side of the robot. The driver actually postioned the shot manually. This year we don't have much for sensors but some limit switches.

If sensors can give us a significant advantage, we use them. If we don't think they will add much we leave them out as they can be just another thing to go wrong in a match.

[billbo911]

Michael,
I for one, truly appreciate the simplicity and reliability approach 1662 has taken over the years. Your success alone is a great testament to it's effectiveness.

That said, if we look at what a "Robot" is: "A robot is a mechanical or virtual artificial agent, usually an electro-mechanical machine that is guided by a computer program or electronic circuitry. Robots can be autonomous, semi-autonomous or remotely controlled...." Wikipedia. Or my "definition": "A electromechanical device that operates and reacts to it's environment based on programming and sensor input.". It seems apparent that sensors are a major part of a "robot". If we neglect to take advantage of their input, are we really teaching robotics.

Please don't get me wrong, robotics is such a huge field that having a focus on the mechanics and less on the sensors is just a valid as any other approach. The last thing I would want to suggest is that 1662 should change their approach.

What I would like to suggest is that you take time in the off season and start experimenting with sensors that you feel you might benefit from. My first suggestion would be to start with quadrature encoders, limit switches, and potentiometers.

Please feel free to contact 2073, EagleForce, if you would like any help with this.