After my last testing, someone more rational than I suggested "he looks pretty good, why not stop with a static droid?" or maybe I "like the mess."
Being less balanced myself, I confess I found myself somewhat at a loss to how to respond to that remark
I guess it's kind of like climbing Everest. I'm at base camp and I'd really like to get to the summit. It may prove impossible (more impossible?) but I've made it this far, so I have to try...
(FWIW, the mess isn't as bad as it looks, "only" 3 of the panels fell off (ok, half the sphere), but the panels stayed mostly intact and it took 20 minutes or so to reassemble.)
For those questioning my sanity, this isn't even the hard part. I'm not at all concerned that I'll be able to get the sphere to roll.
The first roll test showed that the sphere is "round enough" (& stable enough when it gets Kragle'd).
This second test, short though it was, demonstrated that an axle centered on the triangles should be aesthetically pleasing. Three panels fell apart, but without most of the internal structure expected to support the final model that isn't surprising. (He needs more 3D printed panels, and, of course, the Kragle!)
Where I'm afraid the obstacles may become insurmountable is getting the head to smoothly slide over the dome. THAT's going to be the hard part!
I have a fallback plan to make him approachable even if that fails. I also suspect that even if the rolling ball works, he's likely going to be limited by terrain and in speed. L3-G0 placed 3rd in a race, I don't think Lego BB-8 is going to do that!
In my previous post, What's the best way to roll?, I discussed an intriguing alternative to the "hamster" model. That seems like a pretty promising technique, but is a driveshaft going to "look right"? Are the 123/456 triangle centers an acceptable point for the center?
Clearly some orientations are bad. If the panel rings are in the center, then it'd be pretty obvious that two rings were always on the outside while the others chased each other in circles. That might be an effect, but it's not the random looking behavior of the movie.
With the driveshaft ends in the middle of a triangle, it'd kind of alternate which side the panel was traveling up under the dome and look less regular and the motion would seem more naturally organic... In theory at least, but would Lego BB-8 look OK that way?
One way to find out, spin the sphere and see what happens. Unlike the roll test I decided to hold it and make it steadier.
BB-8 says "Ack, More Tests!!!!"
That didn't quite work out.
3 of the panels fell off (that sounds better than saying "half the sphere fell apart"), but the test showed, I think, that this axis of rotation is pretty good. I was going to attempt to rock it a bit as it would when turning, but I think the few seconds of rotation showed that the pattern doesn't look particularly regular, even though it is.
So, it's a go! Depending on alignment with the panels/Lego I might make it a little off-center from the triangle, but I think this will look good enough!
Now that the static model is kinda working and we proved that he can roll, HOW should he roll?
In "How Do I Make Him Drive" I discussed a few options, however since then another has revealed itself. If you watch the Star Wars Celebration: Anaheim BB-8 roll-out video closely, you may notice that he pretty much rolls around the same axis the entire time. That's interesting. There's another way to make a ball roll.
What I think they did (and pointed out by the BB-8 Builder's Club) is to have a "driveshaft" that runs between two points on the sphere. That makes him run in a straight line. To turn, there's a flywheel, likely hanging below the driveshaft as low as possible. Conservation of Angular Momentum means that when the flywheel is spun rapidly in one direction, the sphere has to rotate in the other direction. Together the movement is pretty natural.
BB-8 on Stage at Star Wars Celebration Anaheim
The endpoints of his rotation axis appear to be in the middle, or close to the middle of the white "triangles" between what we builders have labeled the 1-2-3 panel rings and the 4-5-6 rings. (That numbering scheme is a coincidence, I was thinking of dice and the orientation is pure luck, but easy to remember.)
Furthermore, folks had notice an interesting ring on the model where it looks like there's a round access or construction mark. That ring surrounds the 1-2-3 triangle. Not sure if it's related to the driveshaft, but seems likely that there's something interesting to cause it.
Ring on Celebration BB-8 Stage Model
So I think what might work best is to have a "driveshaft" between the two triangle panels. There's a video of model doing this on YouTube.
That technique solves a lot of concerns about my build. The interior was intended to be smooth to allow a "hamster" to roam, but with a driveshaft it doesn't have to be as particular. In addition, it allows for simplified wiring and placement of the LEDs... Eventually maybe even room for some of the toys in the hatch panels?
Anyway, that's what I'll investigate in the future.
Well, since the body shell is being a little bit annoying, I have to think a bit about how he's going to drive. I have a couple ideas and a few fallback plans, but there are a limited number of ways a BB-8 can drive.
Cheerleader
One style that comes to mind is what I call the "Cheerleader" after a video I saw of Murata's cool balancing Cheerleader robots.
In this style, the robot balances on a ball, and can move the ball in different directions by moving the motors underneath it. One of the first BB-8 builders in the UK chose this approach. That makes the "ball" (BB-8's body), pretty simple, but the head is a more complicated robot.
Celebration's reveal of the stage BB-8 showed some of the problems with this approach because BB-8 can lean his head without moving the ball, so the stage droid isn't a "Cheerleader"
Hamster
Another type is what we've generally taken to calling a "Hamster" in the club. Hamster droids basically treat the body like a big hamster ball, with a robotic car hidden in the middle of the droid. This is likely the approach the stage droid took, though there are a few different ways you can build a hamster. Challenges are that the hamster has to be big enough to move the ball, and that you have to be able to open the body to service the hamster.
Hamster Car
The hamster "car" could even literally be an R/C car, driving around in the ball. When the car drove, it would push the ball in that direction. That's reasonably simple, although it would mean that BB-8 would have to do 3-point turns and couldn't pivot. The stage droid showed movements in arbitrary directions, and even swiveled the body, so it probably isn't a car type hamster.
Omniwheel Hamster
An omniwheel drive robot is a typically a two, three or four wheel drive vehicle where the wheels are 120 or 90 degrees to each other. Each wheel also has rollers that allow them to slide in a different direction than the wheel's rotational direction. An omniwheel robot should be able to move in any direction, and can even spin in a circle. A microcontroller sends appropriate signals to each wheel to allow for smooth movement. Here's a short video I found of one on YouTube.
An omniwheel robot inside BB-8's body would allow him to turn in any direction, and even to spin the body in a circle. The hamster could also have arms to keep it centered in the ball and more stable.
The Sphero BB-8 is like a two wheeled omniwheel hamster, and omniwheels are a likely candidate for the stage BB-8 we saw.
Hamster Cage
Another possibility is a frame like one of the gyroscope rides. The outer frame of the gyro would be attached to the droid shell and then motors could swivel the axis to cause the shell to move. In order for this to work, the interior car (where a person would be on a human gyroscope ride) would have to have enough weight to keep it upright. If the center bit was too light, then it'd just flip over instead of moving the ball. Since we need batteries anyway, this isn't a huge problem.
A few builders are considering this approach, however it gets tricky with the frame getting in the way of other effects, like trying to control/attach the head.
A variation would be some sort of track on the interior that allowed omnidirectional movement.
The Power of Attraction
But if you do a hamster type droid, what about the head? Most builders think that the head is attached by magnets to an arm on the hamster. (This is again similar to how the Sphero works). For more complex movements, the arm could pivot to make the head lean in the direction of the movement or to peer around corners. It could also rotate.
A variation of the arm would be to have a car and track mount to the hamster doing basically the same thing. A track would allow better control of the motion since you wouldn't have a 25cm arm with a large with on the end trying to be controlled by a couple motors in the middle. The track could pivot like an Artoo dome, and the magnet attachment could also rotate.
It seems likely that the stage BB-8 is some form of the magnet attraction. Challenges are balancing the magnets, the head still needs power for the lights, and the head needs to be light. Additionally the head has to slide easily on the body, which could be done with round casters.
Combination Mode
Another possibility is a combination of the two: an inner hamster and an independent cheerleader head. They'd have to coordinate their movements, which makes the system more complex in some ways, but they wouldn't have to worry about the mechanical lever arms of the magnetic method. A cheerleader head would probably still have to be lightweight (compared to the hamster) to enable leaning and peering gestures. This is probably more complex than most BB-8 builders need.
What About My Lego BB-8?
The Lego BB-8 has an additional challenge that most builds don't have to worry about: Rough interior & exterior ball surfaces. Omniwheels don't like driving over rough stuff, and an irregular surface makes it hard for the head to slide, and difficult to keep an even magnetic distance.
Because of the rough body, I'm leaning toward a combination hamster + cheerleader. Perhaps even a simple R/C Crawler for the hamster as big rubber tires could easily navigate the rough bricks. The trick is whether or not I can get the cheerleader to drive on the ball.
Fallback plan is a hamster with a magnetically attached head, which probably requires smoothing out at least the inner sphere. Smoothing out the sphere may help with the panel's current structural issues as we may be able to 3D print reasonably smooth plates to tie the panels together and also have a smooth interior.
Well, after some delay getting bricks and life interfering and all that, we've started building... Here's a couple pictures of the building area. Kinda chaotic!
