A few years back, with the pending release of a new Star Wars movie I realised that, properly motivated, I could build my very own astromech droid. Since then I've been mostly learning about tools, and how expensive a collection of them can quickly become...and teaching myself how to do some stuff too. I'm at the point where I'm confident to start, but I've decided it would be much easier if had a bigger printer. However I'm done spending money on tools for a while, so... I've kept my Tomy Omnibot 5402 I've kept since I was a kid, and it's time to give him a 21st century (artoo-like) upgrade. He'll serve the role of a Smart Speaker/Hi-fi/Emulator/Assistant. The plan is to put a Raspberry Pi in it, along with the Google AIY Vision and Voice kits, replace the motors, and modernise the front panel. The problem is I can't quite settle on how to do that so I'm throwing it out to you guys, what options are more fun? Working from the bottom up... Wheels The two options here are to simply replace the motors and leave as is, or add mecanum or omni-wheels. The latter would, I think, suit his character more but it's also more difficult, expensive and probably harder to live with considering the extra rattle. Sensors Beyond the google stuff I plan to put ultrasonic distance sensors on the front and back below the bumper, but are there other sensors you think I should stick in there? Even random things, like do you think he should be able to tell the room temperature Front Panel There are two ways I could go with this one - Either slap a big touchscreen on the front for the minimalist look, or use a variety of different input methods and indicators for a tactile and arguably more retro look. Arms The arms at the moment are very stiff, designed to be twisted in to place then stay there. I could leave them like that, or I could motorise them, which would be cool but really wouldn't add that much utility...or I could straight up replace the arms with actual robot arms. Head The plan is to replace the red lights in the eyes with either neopixels or small tfts - Which? I could also possibly motorise the head so it rotates, but again it's one of those expensive and not-particularly-useful-beyond-looking-cool kinda things. So what do you think? What would you do with him?
Well, if anyone's interested, I popped off the bottom to see what's going on with the motors. And we find this... A self contained little movement system with housing for the motors/gears (along with dust and grime that's been gathering since the 80's). There's nothing particularly interesting looking up the Omnibot's...ummm...midriff - Just a speaker and what I assume is a mic. Above that is the rear of the slide-out tape deck before we meet a black plastic partition Popping off the top of the gearbox... And this is what we find underneath. I'll take apart the rest tomorrow when there's more light, but so far everything looks to be in pretty good nick, beyond a bit of a clean. It's everything I need to know for stage one - Putting new motors/board in it, SR04's, and hooking them up to a Pi3 with the AIY voice HAT. Stage 2 is installing a touchscreen & usb ports on the front. Stage 3 is making the head rotate, adding cameras and replacing the dome (there's a crack) Stage 4 is the final and optional stage - Adding servos to the arms.
Hellooooo, 1980's! I remember those things well, although I never owned one (a Big Trak is the closest I got). I also had a go at a simple pet robot in a science magazine, and have two books on the subject: How to build your own working robot pet --Frank daCosta And Growing up with Lucy --Steve Grand Both are probably a bit more ambitious than you want to go, but interesting reads nonetheless.
Having looked at the synopsis of "How to build your own working robot pet" - that's basically what I want to do. Where I want to go with it is to use it as a test platform for different technologies, be it software like Tensorflow or ROS, or hardware like Lidar or depth sensors (when I can afford it). I couldn't resist and took it apart tonight (Waiting for tomorrow's light was for the camera, not for me). Having inspected everything it looks like even my most ambitious plans will be feasible. I'm still settling on the hardware - no expensive stuff yet - but it should be relatively inexpensive to make my own google spy robot pet.
The technology in the first book is definitely late 70's to early 80's: a (physically) big old 8-bit processor with a bunch of Eproms for the core program (code included in the book) and a large bunch of CMOS logic-gate chips, all wire-wrapped together on a breadboard. Nowadays even an Arduino Uno will probably outperform it; an Arduino Mega definitely. I don't think that the tech will be any use --current stuff is way more powerful and much cheaper-- but the design principles and core functions are useful points of reference. I am strange excited by this project. You have my lathe and my mill.
Also excited by this, I was a 90s child tho so I had this. As for the Lidar option you can get much cheaper static sensors and make your own rotating parts for it.
Well, I've ordered all the parts except the servos. I think I want to get my hands on the AIY vision kit before I start making the thing it's attached to, the head, rotate. Arm movement can also wait until I've figured out where all the new electronics go. While his arms and head won't move yet, he will be able to move around. I sourced motors of exactly the same dimensions of the original, so I should be able to keep the drive train intact, but they're coming from Asia which means they could take up to a month to arrive. By that time, however, I'll have had plenty of time to plan out the animatronic aspect, so I'll be installing the servos then too. We're looking at February/March before he's built and adequately coded.