/robowaifu/ - DIY Robot Wives

Advancing robotics to a point where anime catgrill meidos in tiny miniskirts are a reality.

Canary has been updated.

Build Back Better

Sorry for the delays in the BBB plan. An update will be issued in the thread soon. -r

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB


(used to delete files and postings)

Have a nice day, Anon!

Reploid thread Robowaifu Technician 02/28/2022 (Mon) 04:13:32 No.15349
A few people thought it'd be a good idea to start a thread for Reploid builds, so here we are! To kick things off, here's a little progress on painting my own RiCO. It's just spray paint so it doesn't look stellar and I screwed up a couple parts. All the blue trim paint needs done as well. I don't care if it's perfect, I just want to get her done. Not to mention anything adjacent to "art" or "craftsmanship" is beyond me, mostly due to lack of patience: I don't want to spend 1000s of hours grinding away with a paintbrush when I could be designing up cool (to me...) robotic mechanisms for instance. I bet you bottom dollar none of my projects will be winning awards in the fit-and-finish department. Can't wait to see what happens with Pandora and whatever other Reploid projects people might be working on.
>>15349 Really cool idea for a thread. I've read up on Reploids from the day of the Mighty Number 9 (sad that the game didn't turn out well, really liked Call's design!) kickstarter to get more familiar with Megaman. I think its a very good idea to base designs and features on the Reploids from Megaman like their soft exterior design and how they're built to resemble humans with "armor" on. How large is that Reploid kit that you have? Can you put servos in it and make it a base for your robomaid or is it just large enough to be a figurine?
I love the smooth shiny finish you got there with the resin prints. Makes it look factory made. With FDM it is rather a chore to get parts looking even half that shiny (much sanding and gloss paint or epoxy resin).
Glad to see the new thread OP, thanks! :^)
>>15350 She will be somewhere around 500-600mm tall I think. Hard to tell without the legs even designed. >figurine Where we're going we don't need no static figurines. She has 7 degrees of freedom in her head/neck and 4 in each arm so far. I want to add 1 per hand and 2 or 3 in the waist, maybe wheels for mobility later. >>15358 You asked about a tutorial in the old thread, I haven't even finished her body yet so a tutorial is a ways away. That said if there's something specific you want to know about just ask.
>>15362 How did you do the neck and arms? Had my own ideas for everything else.
Unfortunately painting is soooo much fucking easier said than done. After spending a few days trying to do the beginning of the paint job and failing miserly, I culminated in this. Blue detailing sprayed fucking everywhere. Holes in the paint, "fish eyes", peeling, surface finish of a clay pot from 5000 B.C., you name it. This was from the last moments before I pitched it all at the floor in a fit. Needless to say RiCO probably isn't happening any time soon. Almost everything has to be remade from scratch. And if I ever do, it's only ever going to be in fucking ugly ass 3d-print gray.
Open file (178.67 KB 1260x880 cool art.jpg)
Open file (100.61 KB 958x604 Rico every 1-5 levels.jpg)
>>15368 >Needless to say RiCO probably isn't happening any time soon. Almost everything has to be remade from scratch. And if I ever do, it's only ever going to be in fucking ugly ass 3d-print gray. Did you even prime the parts first? If you aren't going to use these prints then it would be helpful to share how you made all the screw holes and connector parts to speed up making similar themed projects with our own work. >Holes in the paint because that's how that happens. Had thought of printing in several different colors like Gunpla myself but a whole different roll for different parts of each color would not be as economical as the goal of budget DIY robowaifus would want. On another note: How did you do the neck and arm internals? Had done some rough sketches of a skeleton since I hadn't seen that much from Kiwi and now that I have another chance to get back to modeling, I can start the Pandora thread within the next few days and use this as a general collaborative thread for organization purposes. I have no nice things to say about the cyan thing /s. Not a bad character, but the unfortunate face of the epitome of what a tutorial should not be. She's constantly still assuming control away from returning players to this day. Pic Rel is the mood in my own travels since we're doing the whole adventurer thing and I like the picture.
Edited last time by AllieDev on 03/02/2022 (Wed) 04:28:16.
>>15373 >paint primer Yeah I'm sure that had a part to do with it. "Self priming paint" my ass. Sorry, I really lost my cool over this earlier today, and destroyed most of the SLA parts in the process (today has NOT been my day). Due to my lack of dedicated work area I'm unable to install a paint booth or house a lot of painting supplies, so I'm limited to whatever comes in a spray can. Hand painting is out of the question, as I mentioned earlier I have zero skill when it comes to handiwork and zero motivation to drop everything and practice for an extended period of time. >arms Please check out this STEP file, do note it was pieced together from several smaller working files so the arm->body dimension might not be quite right. Additionally, the face mesh is missing, if you really need it I could probably get STLs but not much else. Otherwise this should be a complete assembly of all RiCO hardware designed to date. https://files.catbox.moe/rhui45.gz >RiCO as a character Funnily enough I've never played the game she's from, mobile gacha casinos aren't my jam. She was floated as a possible IRL fembot design by someone else, it was a reasonably practical design and cute, I was sold.
>>15368 >Unfortunately painting is soooo much fucking easier said than done. I'm an artist and acrylic painter myself and plan on hiring someone else to paint my parts. Not only does it take skill but also mind-numbing patience and attention, and it doesn't make sense to do a $15/hr job yourself or spend even a 100 hours training yourself when you can do something else that pays much more than that. I hope you don't get too discouraged by the setback. Your design is looking really good so far and I'm looking forward to seeing the mechanisms you come up with for the torso, legs and hands. There are lots of people who offer affordable 3d printing and painting services on Fiverr and probably other sites and forums too if you message them and ask. If anyone else attempts to paint their prints though, I don't know which primers are ideal for different materials but paints don't stick well at all to plastic or resin. A primer is mandatory for a good paint job, and resin needs to be cleaned thoroughly for even primer to stick. Some say they prefer using spray primer for resin prints since it's easy and adheres better, some prefer gesso since it can be sculpted and refined. I prefer using gesso since I don't have much space or ventilation. Ideally 2-3 coats of gesso should be put on first as thin as possible. It will shrink as it dries and may cause holes and patches to form where put too thin but too thick will not adhere well. If the gesso is uneven it can be easily sanded down with a fine grit to get a smoother finish after painting. Hand-painted acrylic has to be applied evenly and ideally in one go to avoid creating a bumpy surface, which is really difficult if you're working in a dry environment. Acrylic paints can be mixed with a slow drying medium to give more time to work with them and create smooth gradients without airbrushing. If you make mistakes though they can be fixed by sanding down with clean sandpaper. Masking fluid can be used to avoid painting over already painted areas and safely peeled off later. Lastly, the paint can be varnished with a clean gloss, satin or matte finish that will also protect the paint from damage and marking, ideally using a UV protective one to protect the colors from fading too. I might've missed something but these are the most important things to know to achieve a clean finished look. I'd be interested in doing a 3D model of Alia if anyone is interested in building her or parts for RiCO if you need any. I really love their designs.
>>15368 There can be beauty in slight hand-made imperfections anon. Wabi-sabi and all that? However, your story is all too familiar to me anon. Even after many hours designing, the frustrations I have had developing Sophie have been so great at times that I had to just cool off and work on something else that my creativity was telling me to do. Don't be too hard on yourself though. I have parts which I know are currently pretty crappy but I also know they can be improved in time.
>>15362 >That said if there's something specific you want to know about just ask. Well, everything ofc! -I'm only vaguely familiar with resin printing. All about that would be good. -How you worked out the design process. -How did you create the mold for the silicone, then pour it/fashion it. -More about how you plan to articulate her. And anything else you'd like to share with us here. We have a whole thread, right? :^)
>>15388 >I'm only vaguely familiar with resin printing. All about that would be good. There are probably people out there that have explained everything you'd want to know way better than I could. Stretch your google-fu. For designing parts to be printed there are a few major things that I like to keep watch of: - The orientation of any critically flat surfaces, such as mating surfaces. These must be facing away from the build plate, or else sagging artifacts will be present. - Whether or not the part will have cause "cupping" when printed at a said orientation - If you try to print an upside-down bowl, or any geometry approximating one, on an SLA machine, you create a sealed pressure chamber that can cause the print to fail. Vent holes may need to be added, or other changes made to avoid the "cup". - Generally what orientation to print the part in. This is determined by many things: The positioning of mating surfaces or functional features (want to be kept away from supports at a positive angle), what areas need good surface finish (these also need to be kept away from supports, unless you want to spend all day sanding), how many separate "islands" the part will form on the printer as (if you print a "V" shape with the top towards the build plate, the 2 sides of the V are separate parts until the end of the print. This mating will cause some amount of deformation, and small extremities printed this way might fail altogether), the print time, and more. SLA is not ideal for mechanical parts, and it mostly comes down to trial and error. >How you worked out the design process. I believe I started by importing a reference drawing of RiCO as well as a mesh someone made into Fusion360. Then I loaded in a 9G servo model and laid out where all the servos could fit. Once I knew what joints I could have and where the motors could go, I more or less modeled her from the outside in - first creating a solid form, then hollowing it out to create a clamshell and adding the pivot joints, servo mounts, screw bosses, and other internal features. These parts were inspired by how the injection molded housings for power tools and small appliances are designed. This relatively primitive design doesn't have bearings for each joint, only plastic-on-plastic fits. They will wear out eventually, but this was designed more as a "first build" project, something cheap(er) to put together. If you do build it, put some grease on the joints to prevent early destruction. The cheap RC servos I picked will almost certainly break first (and they don't have real bearings either). I really encourage you to load up the CAD file I posted and look through it, a 3D model is worth a million words. >How did you create the mold for the silicone, then pour it/fashion it. I posted some details about the material on the old thread, check that out. I took the mesh file for RiCO (the head is the only place I used an existing mesh model instead of re-drawing in CAD) and, in Blender, created a 2 part mold like you see here. I don't recall the exact process I used to create the mold, however I do remember it being tedious. A lot of use of the "boolean" modifier. Note the pink pieces and how they're covered in cones that "lock in" to the molded silicone, these pieces have screw holes on the other side to allow for easy mounting to the main assembly. Everything here was FDM printed except the gray piece which was SLA for a smooth finish. >More about how you plan to articulate her. For everything I've completed look at the CAD model, it should be self explanatory how it works. For the non-completed parts: the waist I plan on having a 2 or 3 axis tilt mechanism, similar to the neck but stronger to handle all the weight, and surrounded in a silicone casing to mach the reference. I don't even know if it could fit, I haven't gotten to designing it yet. Any plans beyond that are just ideas at this point. For electronics: Motion control is accomplished via ESP32 (mainboard with camera mounted in the chest) receiving commands from a host PC via TCP over wifi. The commands are stored in a buffer. When a special playback command is sent, the last command from the buffer is converted from direction-velocity to a sweeping PWM signal sent to a pair of PCA9685 servo drivers. Simultaneously, another thread on the ESP32 is broadcasting a camera feed back so the host PC could do object/person tracking to facilitate user interaction. (Note no PC software has been written yet!)
>>15390 >using ripped models They’re high quality models, but made to be game assets first and foremost. Was better to use them as size references since there’s no real reference sheet for most of the x dive iterations of the characters. How did you go about the bore threading on rico?
>>15392 >How did you go about the bore threading on rico? I used self tapping plastic screws, however I'd recommend using an M3 tap instead (resin takes to cut threads better than soft FDM plastic).
>>15394 >M3 tap Any specific brand for the bit you would recommend? I was planning on trying out the self tapping screws first and just not overthinking the bores anymore. Only concern I have is how likely are they to damage the resin/plastic filament past the useful initial stage of making the bores in the first place when removed and reinserted? >For electronics: Motion control is accomplished via ESP32 (mainboard with camera mounted in the chest) receiving commands from a host PC via TCP over wifi. The commands are stored in a buffer. When a special playback command is sent, the last command from the buffer is converted from direction-velocity to a sweeping PWM signal sent to a pair of PCA9685 servo drivers. Simultaneously, another thread on the ESP32 is broadcasting a camera feed back so the host PC could do object/person tracking to facilitate user interaction. (Note no PC software has been written yet!) On the topic of electronics, have you put any thought into tracking and blinking eyes? Not much of a fan of the chest camera and would rather put the cameras inside the eyes, but it does make some sense.
>>15402 >Any specific brand for the bit you would recommend? I was planning on trying out the self tapping screws first and just not overthinking the bores anymore. It's plastic, you'd be hard pressed to break a tap, quality doesn't really matter. Just make sure you get a normal cutting tap and not a form tap. Make sure you have the right drill size on hand too.
Open file (51.70 KB 460x996 AliaCute.jpg)
Rooting for you RiCO Anon. Your high quality results and clarity of information are both deeply appreciated. >>15376 Your design is clear and easy to understand though, sharing so much through catbox is great. Though, much of her design only works at her small scale. (Just a note for fellow Anon's who may want to upscale her, significant changes would need to be made.) >>15377 If you model Alia, I'll build her after Kiwi/Pandora/Madoka.mi. Beauty like hers should be made real. >>15392 >Threading It's actually really easy with plastic prints. All screws are self tapping in most printed materials like PLA or PETG. Just make a hole slightly smaller then your screws major diameter. (The outer diameter of the threads.) This link provides further relevant information. https://www.fictiv.com/articles/how-to-choose-the-best-fasteners-for-3d-printed-parts It is worth mentioning that my method is technically one of the weakest ways of screwing prints together and should not be used if you plan of unscrewing often. It's good enough for almost anything and I've had lots of look with it.
>>15416 > All screws are self tapping in most printed materials This is true for FDM materials but not so much for SLA parts, if you try to force in an M3 it will often split the area around the hole. To get a self tap into resin without breaking the part you have to make the hole size barely smaller than the outer thread diameter, giving you a rather weak bond.
>>15420 Thanks, that's good to know.
>>15390 Thanks for such a detailed and thoughtful effort-post response Anon! This kind of information really helps us all out. I'm glad to know that you have experience with both SLA printing and with silicone molding. We'll all need proficiency with these skills before all is said and done IMO. Interesting about the potential of a vacuum seal forming a 'cup' during a print. Makes sense, but it's something I probably wouldn't have foreseen in advance. Thanks. I did grab a copy of the file, but I don't currently have a CAD program. I wonder if there's some way to open it inside Blender? I really like the compact fashion you've integrated the shell, actuators, electronics. Keep going Anon! Cheers
And we're back in business! All the resin parts have been replaced.
>>15839 Oh wow, she's adorable. Great work, anon. Will her eyes be able to move? What's the range of motion on her arms?
>>15839 Truly gorgeous work, I'm getting a SLA printer which will arrive on Tuesday. FDM printing just has countless problems when it comes to dimensional accuracy and repeatability. Thanks for the advice from your previous posts, hope to catch up to you soon. The ball joint on her shoulders is pretty clever after another look.
>>15842 >SLA 3D printer >Accuracy Prepare to be disappointed. It's way easier to get good accuracy off FDM. I had to drill, sand, or file off something on every functional feature on this. FDM beats SLA for prismatic geometry and anything involving accurate hole spacing, sizing, press or slip fits... all the mechanical stuff. SLA loves to warp and distort your parts in lots of interesting ways. SLA is for organic "curvy" parts, good surface finishes and odd geometry you couldn't print very well on a filament printer. Case in point: all Rico's internal parts are filament prints. If I had to do those complex sliding mechanisms in the head in resin I'd be fighting it to this day. One thing the resin has going for it is it's way easier to cut. Thermoplastic turns to a gummy mess the moment it sees a blade, drill bit or sandpaper. SLA resin fairs much better in this regard, easily abrading away into a powder. Provided you're careful, as unlike traditional plastic it has a tendency to split/shatter under too much load. Either something's up with your 3D printer, your parts aren't designed around 3D printing's limitations, or you need to turn down your expectations a bit for tight fitting parts. It's still nice to have resin printing available for outer shells and other "look good" parts, congrats on your new machine! What did you get? >>15840 As I mentioned above she has 7 DoF in the head, 4 in each arm, with hand and waist motion planned for the future. Specifically, the head motion is as follows: - head rotate left/right - head rotate up/down - head tilt - eyes up/down - eyes left/right - eyelids close - mouth open Each arm has 2 shoulder joints, an elbow joint and a wrist joint. I'd like to get a demo of at least hear head moving sometime, but it'll probably be a while.
Open file (148.74 KB 1600x1600 Mono4k.jpg)
>>15843 I'll try to keep my expectations in check. I bought an Anycubic Photon Mono 4K. My old Ender 3 has terrible accuracy with mechanical parts so, it's hard to imagine the new one being worse. I design everything around my printers limitations but, those limitations are rather well, limiting and the printer gets worse over time. Could you provide some examples of warping behavior to be aware of? I know there's significant warping on any part that touches the build plate, the dreaded "elephants foot." I thought that for everything else you could get sub mm tolerances but, that's from reviewers, real world examples are more valuable.
>>15844 Here's a simple example of filament vs resin issues. The blue part has 2 M3 counterbores and a mating feature that fit into the green part. The triangular extrusion on the blue part needs to be in contact with the corresponding cutout on the green part. (Imagine it's a load bearing or locating feature). This would be a walk in the park to do on an FDM 3D printer. Both parts lay flat on the build plate, meaning that any inaccuracies in the machine will be applied to the mating features in both parts the same way. I put .3mm of clearance between the triangle on the blue part and the hole on the green part. I know from experience that this is about right for my machine; if it's too big or small I can easily find a good fit by trial and error, as the cost per part is about $0.50. The only other possible issue would be the counterbores on the blue part, it may need supports. No biggie. Doing this on SLA is much harder. Note that you can't just plonk down the part onto the build plate anymore, now it has to hang from a support tree. One of the many limitations of this is that you cannot have a face parallel to the build plate on the plate side; everything must be at an angle. Also: - No sudden changes in amount of material per layer can take place. - The print should ideally have a single origin point where the model starts from: think of a tree, starting with the trunk end growing out towards the branches. If you print the branches first, you effectively have multiple separate parts that must meet during the printing process. These will be misaligned and can fall off. - You do not want a part that has solid area across the majority of your build plate, or might be too heavy. It could fall off the support tree. Think that's all? No wait there's more! Remember, we're dealing with a vat of liquid here. That means a few things: - The part is raised up and down into the vat for each layer. Because of this, if we have any concave features facing the printing vat, the part may be ruined because the hydraulic pressure/vacuum formed as the build plate travels up and down. This is sometimes called "cupping" in resin printing circles. Concavities must either be avoided or holes leading to the plate side of the model must be added to the part. - every time the printer pulls off another layer from the vat, a force is applied to the model. For simplicity sake, imagine this as a force normal to the build plate, pulling any small features that stick out horizontally down towards the vat. If your support structure isn't bulletproof, your part will be partially lost. As you can see, making something that works on SLA is a *bit* more involved than FDM. Thankfully, there are tools to help us - auto orientation and auto support. But you still need to keep everything I mentioned above and more in mind to double check and adjust as needed, as the auto tools only get so far. See how I had to position the blue and green part in the 5th picture (resin slicer) vs the 4th (FDM). This will probably work, but there will be several issues: - The faces where all the supports are attached will be anything but flat. - If we tried to orient the parts so that the mating faces are coplanar, the print would likely fail as the cutout for the triangle in the "L" part would form a "cup", tearing the part right off the support tree. - Because we can't make the mating faces coplanar, they will likely be distorted by all the forces at play, causing them to not mate well. I can tell you from experience, the faces won't even be flat. The only way to make this work will be to manually sand down the triangular feature by hand until a good fit is achieved. It not fit nearly as well as the FDM parts will without sanding. - Notice how the holes on the "L" part are running parallel to the build plate? Just like in FDM, holes made this way will not be well formed. So count on drilling those out manually. The counterbores will also need some help. - If you're not happy with the fit and want to re-print with adjustments, that'll be $4 in resin, plus gloves, plus IPA, plus an hour of your time processing the part. It adds up fast! I hope this example helps explain why SLA is more of a PITA to get good parts off of.
>>15839 A cute! Nice character appeal going there Anon. >>15842 >I'm getting a SLA printer which will arrive on Tuesday. Gratz Kiwi! >>15845 Outstanding! Excellent breakdown, RiCOdev.
Open file (38.30 KB 1029x800 pg1.png)
Open file (47.12 KB 1028x795 pg2.png)
Wiring diagram for those playing along at home. It's really simple but some people may be new to this stuff and want a guide for a similar "doll scale" build. Do note if you use this configuration, the battery is *NOT* protected from your external 12V supply, it's up to you to turn the disconnect Off before using external power. (You may want to place a diode after SW1 to prevent this issue.)
>>15977 Thanks for sharing! For those interested, his circuit from: https://www.bitbuilt.net/forums/index.php?threads/portables-and-batteries-a-guide-and-explanation.2228/ is good for basic battery protection. (some battery charging controllers have this functionality built in)
>>15977 >>15979 Thanks Anons.
Open file (17.07 MB 840x756 0295-3349.mp4)
Here's a short clip of me jogging her around, still have a lot of work to do though.
>>15999 She is looking really good so far, RiCOdev! I'm personally quite interested to see how you plan to manage motion control in particular, as she grows more complex over time. Regardless, keep up the good work Anon! :^) Cheers.
>>15999 That is impressive. How did you go about making the head tilt like that and blink? It would be great to have some demonstration of how to install the servos and new components to the main board for the first time. I am thinking of getting an esp-32 CAM for the microsd card slot since I’d already picked one up for pandora months ago.
Open file (106.86 KB 769x720 ClipboardImage.png)
>>16007 >That is impressive. How did you go about making the head tilt like that and blink? The eyelids are just paper glued to a roughly U-shaped part that pivots at the back of the head. The neck joints... are just servos I guess?? idk how to explain
>>15999 She's really cute! >>16018 I believe you've implemented a pan tilt mechanism. Hope that can help Anons. >Paper eyelids Great idea, mind if I borrow this technique for MaidCom?
>>16018 Nice design Anon!
>>16021 >Great idea, mind if I borrow this technique for MaidCom? Please do. Check out how I did the rest of her eyes as well, perhaps it could scale up.
>>16029 It does scale up really well. Though I will innovate upon it for Pandora.
Edited last time by AllieDev on 04/27/2022 (Wed) 01:32:08.
>>15999 I'm really impressed with this. I have a lot of questions, I'll scan the rest of this thread to make sure they aren't already answered first.
Curious if you had any changes this week with your wonderful robowaifu, RiCOdev?
>>16243 RiCO will be on hold for the near future; your friendly neighborhood reploid builder has got bigger fish to fry. I'll still keep an eye out here to answer questions etc.

Report/Delete/Moderation Forms