linor
New Poster
Posts: 7
|
Post by linor on Jun 15, 2020 14:19:58 GMT
As one can never have enough droids, I ended up thinking which droid I would make next. As I try to create less common droids to add to the diversity at events (I own a C2-B5 unit for instance), there are still plenty to choose from.Having played Star Wars: The Old Republic myself, and some rumours that the movies might be exploring that time as well, T7 ended up high on the list. T7 however, is not a droid without challenges. In this thread, I will try to post some of the more relevant choices I had to make, and why I selected that particular option. This won't be a full build log, partially because I know I will forget to update it, but also because I feel T7 is an 'experienced builders' droid. There's no existing parts someone can sell you and no 3D model you can just print. The choices made by me are also based on the tools and materials I have access to.
Disclaimer: While I have no problems sharing things like CAD models or Arduino codes, do not expect any form of support on them. They are provided as is. I expect that they won't immediately work for your situation and will require tweaking by you to get it to work.
|
|
linor
New Poster
Posts: 7
|
Post by linor on Jun 15, 2020 14:29:12 GMT
Starting point
With the choice for T7 made, I tried to determine the things I wanted to accomplish in this build. They are, in no particular order:
* Reuse the Q85 motors and VESC controllers that I was not using in my R2 build. * T7 needs to fit in my car, so it needs to be possible to be taken apart into manageable shapes * It needs to be possible to drive T7 in locations with less then smooth floors (concrete with holes, outside pavement etc) * 3D prints only for details, given the need for disassembly, and because I do not believe in 3D prints for structural items. * Convention proof, so few openable panels etc
|
|
linor
New Poster
Posts: 7
|
Post by linor on Jun 15, 2020 14:40:43 GMT
Challenge 1: The modelWith T7 being a computer game character, there was plenty of visual material available, with the ability to watch the droids from many angles in the SW:TOR game itself. However, 3D game models can hardly be used directly. Not only is the sizing wrong, it's usually also a low polygon model and details are added with textures and bump maps instead of actually being modelled in. So, the first part of building this droid was to establish how I wanted the droid to look, what details to add and to make sure it was of the correct dimensions. I imported the STL file of the droid into Fusion 360 and then basically redrew the entire droid from scratch, using the in game model for scale and the 3D model at p3d.in/SERep?fbclid=IwAR2n8S0Jb4XIeEcucENwNLbxJSp-wZ8WLxhLmlP2AmuTmP6Pq0j5ZILZ7Gk for reference, along with a variety of reference shots extracted by fellow droid builder Roel, which separated the in game model into parts like legs, head etc. I ended up with the following model: While not useable in it's current form (intersecting geometry and solid parts), it serves as a guide for the rest of my build. Attachments:
|
|
linor
New Poster
Posts: 7
|
Post by linor on Jun 15, 2020 14:55:11 GMT
Challenge 2: ScaleScale of the droid is an issue. On the one hand there's the in game footage and concept art to give some kind of scale: This is my favourite shot of T7, and a setup I eventually want to recreate, including the armoured jedi costume. But that's a challenge for later. * This concept art clearly this shows T7 to be about waist height, and given that I am 1.90m tall, that's quite a bit. * Then there's the wookiepedia page on T7 that lists him as 1.15m tall. * Then you have the 1.10m length of a 3 legged R2 unit * And finally there's the problem of T7 being a very wide droid, wider then an R-Series droid, which leads to issues at doorways. That last point is definately an issues, and it originally caused me to scale T7 to a width that would fit an 80cm door (resulting in a 93,5cm tall droid). After some renderings however (thanks again Roel), it became clear that the smaller version was not an option: So after reviewing the situations/places I would end up using him, and consulting with some other builders, I decided on making him 1.15cm tall; that however made the ability to disassemble him for transport even more important.
|
|
linor
New Poster
Posts: 7
|
T7-O1
Jun 15, 2020 15:13:12 GMT
Post by linor on Jun 15, 2020 15:13:12 GMT
Challenge 3: Driving and the rear wheelI've been driving R2 units for some time now, and I am all to aware of the stress the foot setup with the drive system in the rear causes on the frame. It also limits the obstacles R2 can overcome and it's ability to drive on non flat surfaces. As I had some spare Q85 motors with off-road tires, which I really wanted to reuse, I wanted to recreate the more ideal robot setup; drive system in front, and caster in the rear. BUT, T7 does ofcourse not use a caster wheel, but rather a massive wheel that just pivots around it's axle. So, I needed to figure out if I could make this work, and, rather then just creating the entire droid and then working out the rear wheel situation, I decided to create a prototype to test this functionality. So after some CNC-ing and 3D printing, I had myself an accurately sized setup with which I could use to try the driving system. Unfortunately, initial tests with a passively swiveleld rear wheel quickly showed that that was not a viable solution. The wheel did not function as a caster due to it's pivot location being in the center of the wheel, rather than it being dragged into the correct position like with real casters. So, it was back to the drawing board, trying to come up with a solution for actively turning the rear wheel in the correct direction.
|
|
linor
New Poster
Posts: 7
|
T7-O1
Jun 17, 2020 11:21:08 GMT
Post by linor on Jun 17, 2020 11:21:08 GMT
Seeing that the passive solution did not work, I decided to actively try to turn the rear wheel into the correct direction. As this was still a prototype, I designed a simple bracket for a servo that I could attach to the bearings inside the rear wheel. This enabled to me to adjust the angle. Unfortunately not the whole 90 degrees like in the game
|
|
linor
New Poster
Posts: 7
|
Post by linor on Jun 17, 2020 12:20:51 GMT
Ofcourse the issue with the servo is, is that you need to know the angle at which to point the wheel. After some digging into 'Differential Drive' systems (the name for steering by adjusting one wheel's speed), I found out that it's possible to calculate the Instantaneous Center of Curvature, basically the point around which the droid is turning. Instrumental in this was the paper I found at www.cs.columbia.edu/~allen/F17/NOTES/icckinematics.pdf . Combined with the ability to read the RPM's from my motor controllers, it was possible to use the formula in this paper to calculate the ICC. From there on it was a matter of some simple trigonometry (although I had to dig deep to recall my highschool material ) The VESC returns ERPM's instead of RPM's, which is basically RPM * nr of motor poles * gear ratio. I ended up guessing this value based on some simple tests (set speed at somewhat steady ERPM and count the nr of revolutions in a minute). It seems quite accurate. Oliver mentioned on Facebook that he recalled the motors being 20 pole, so with our 100 value, that would give a gear ratio of 1:5, which does not sound all that weird. The end result was that the wheel actually moves quite well when driving. As indicated in the paper however, when one wheel is not moving, the ICC is at l/2, or about 320mm in the case of T7, which would result in a rear wheel angle of 70 degrees. As I limited the angle in software at 25 degrees maximum, and it currently has a physical limitation of 30 degrees, it basically means that turning that way is not possible (let alone driving one wheel in the opposite direction of the other). I will need to adjust my mixing routine to take this into account.
|
|
tbou
New Poster
Posts: 5
|
T7-O1
Nov 24, 2020 22:48:37 GMT
via mobile
Post by tbou on Nov 24, 2020 22:48:37 GMT
This is quite awesome! How is it looking now?
|
|
|
T7-O1
Dec 21, 2020 8:59:00 GMT
Post by RobHowdle on Dec 21, 2020 8:59:00 GMT
This is quite awesome! How is it looking now? From what I recall, he has been working on the rear wheel building a custom one mostly to get a larger area of turn I think. So far it looks very promising!
|
|
|
T7-O1
May 13, 2023 4:45:12 GMT
Post by zahcstarkiller on May 13, 2023 4:45:12 GMT
Extremely cool stuff! Have you made any progress lately?
|
|