Has anyone tried building FRC waldos lately?

[gblake]

Waldos/Waldoes = Heinlein reference (https://en.wikipedia.org/wiki/Remote_manipulator)

The devices and software needed on the human side of the interfaces have been steadily getting smaller, more accurate, easier to use, and *cheaper*.
YouTube video of an example Full arm, plus hand and fingers mimicked by the movements of a (software-only) virtual person (https://www.youtube.com/watch?v=5ESsEXsImuc).

I'm curious if this sort of thing (which has been tried in the past, but was typically a novelty and not at all mainstream) is becoming more common in FRC robots, either as a way to steer/drive the robot, or as a way to control "manipulators".

Has using these sorts of human-machine integrations become the default for any teams?

For teams that have tried creating waldo-style FRC bots, but have now stopped, what were the reasons for stopping?

Blake
PS: I realize that FRC bots *don't* have have a zillion little motors to use mimicking the dexterity of a human hand. I'm thinking about how the motors that are on the robot, however few that might be, are controlled.

[feverittm]

We used a 'waldo' (and yes we called it the Waldo) in 2011 to control the main 2-jointed arm we used to grab tubes. We liked the way it offered direct control. We were considering using one for a manual control box this year to control the height of our elevator.

However, it can cause some extra work to use with automated systems (like a button to move to a pre-programmed location as we might have used this year to move an elevator to a precise tote/bin height). The issue is when you transition from waldo desired position to/from automated position. You might have to used motorized joints in your waldo (like a motorized slide poteniometer) to track position between automatic and manual operations.

A waldo is a cool idea, however it is not perfect for every game. I can see that for games that use moving actions (like arms/elevators) it would be useful, while for others (like the shooters form last year) it might not be as useful.

(BTW: We still have our waldo in it's nice storage box in our lab, while the robot it controlled has long since been recycled)

[GeeTwo]

I've read the novella several times, and thought about it a good bit. Using a full Waldo input device really only makes sense when the output simulates the human hand/arm (or other analogous limbs) to a reasonable degree. The human hand and wrist has at least 20 degrees of freedom, and FRC robots usually have somewhere between two (for a kitbot with a box) and the low teens (for the robot of a legendary powerhouse team with swerve drive and a couple of rather complex manipulators). Another feature of the true Waldo that cannot be simulated by most of the input glove devices presently available is the feedback pressure. A true Waldo would require at least thirteen analog feedback devices which would provide pressure to each finger tip, each finger inner joint, thumb opposition, wrist rotation, and wrist flex; about twenty would be even better. I believe that most competitive robots at the regional level have fewer than ten independent manipulators and no more than six sensors. Team 3946 has had as few as three independent manipulators (2015 Recycle Rush) to as many as eight (2012 Rebound Rumble: 3 drive (includes shift), 4 pickup/aim/shoot, 1 bridge; or 2013 Ultimate Ascent: 2 drive, 3 climb, 3 shooter*). None of our robots have had more than three truly "analog" manipulators; most of our manipulators are binary (on/off) or trinary (forward, off, reverse). Given this, a multi-button joystick, xBox controller, or the Wii controller seems to fit like a glove better than a glove.

* We had two shooter motors that ran at different speeds, but the first always ran at ~70% of the speed of the second, so I consider these two motors as one independent degree of freedom, especially as regards the driver controls, which simply allowed the driver to turn both on or both off.