11/02/2021 at 17:50 •
I was recently impressed by the work of a finalist in "Redefine Robotics". The robot "Tardygrade" is made up of mostly interlocking parts that fit together like a puzzle, and uses very few screws. Stack-chan, on the other hand, uses up to 24 screws in total. I wondered if I could make this a little easier for young hackers who are starting to develop robots.
One part of Stack-chan's exterior that requires a particularly large amount of screwing is the fixing of the servo motors. This was based on the way that humanoid robot kits on the market are fastened, but I figured that for a small body like Stack-chan to have pan-tilt movement, it would not need such a strong fastening. So instead of using screws to fix the servomotor, I decided to use a structure where the servomotor itself is fixed by snapping it into the frame.
The following is the bracket that I improved after more than a dozen attempts.
To make the servo motor snap into place, the claw part of the bracket is shaped like a seat with curved corners, like the Fritz Hansen Seven Chair. This structure makes it possible to open the claw and insert the servo motor without breaking the bracket, even when using a rigid filament such as PLA. This means that the four screw tightening per motor is reduced to zero.
*Note that when printing with FDM, the stacking direction should be perpendicular to the curve.
I have also improved the fixing of the servo horn and shell. The last version used three or more screws, but by adding a hole on the shell side that meshes with the horn's protrusion, it can now be fixed with a single screw.
Now let's compare the old (left) and new (right) Stack-chan.
Through these improvements, I finally able reduced the number of screws tightened from 24 to a minimum of 6!
I included a number of other minor modifications in this version, such as eliminating the gap between the M5Stack and the shell, and making the hole for the Grove connector on the top of the body opt-in (which can be cut out later with nippers). One of the improvements that developers will love is the Lego-compatible holes on the side of the body, as many modules in the M5Stack series use Lego Technic-compatible pins. This hole will also help to fix the official sensor unit. And of course, the Lego bricks!
You can view the STL and STEP data in the GitHub branch, which I will be merging into the main branch soon!
10/29/2021 at 08:00 •
紅樹 タカオ (@mongonta555) is an expert in both software and hardware and is on his way to building a little fluffy robot park.
He was inspired by tweet from @robo8080 to start building cat-ear options for Stack-chan.
猫耳のパーツ化、、、かなり難しい(;´･ω･) pic.twitter.com/6TwwlJaIvg— 紅樹 タカオ💉💉 (@mongonta555) August" class="redactor-linkify-object">https://twitter.com/mongonta555/status/1425748196060385284?ref_src=twsrc%5Etfw">August 12, 2021
At first, the design was difficult, but thanks to his openness about the process, he was able to get advice from other makers and gradually refined it.
August" class="redactor-linkify-object">https://twitter.com/mongonta555/status/1425830406230708235?ref_src=twsrc%5Etfw">August 12, 2021
In the end, he finally succeeded in creating a removable cat ear option!
October" class="redactor-linkify-object">https://twitter.com/mongonta555/status/1447858958182453253?ref_src=twsrc%5Etfw">October 12, 2021
After ears, guess what's next? Yes, tails! He even went on to make an extra bracket for the tail!
October" class="redactor-linkify-object">https://twitter.com/mongonta555/status/1450443251899461634?ref_src=twsrc%5Etfw">October 19, 2021
The tail is equipped on top of a third servo motor, so it moves. Wow.
Since I made the 3D data available early on in the development of Stack-chan, many community members have been modifying the case and adding optional parts like his works.
Among them, He contributed by creating a case that is compatible with the official module of the M5Stack, which is used in the example above. Although it takes up some extra space compared to the dedicated Stack-chan board, the ability to divert the official module makes it even easier to assemble, making it a useful reference for other community members.
His modified version of the data is available as a pull request in the repository (to be merged in the future).
10/28/2021 at 14:35 •
robo8080 (@robot8080) is known to be one of the biggest fans/longest involved contributors in the Stack-chan community since I released the M5Stack-Avatar library, the predecessor of the Stack-chan project.
He has been exploring the path of "cute robots" such as face customization and integration with various web services especially voice assistants for several years, The following are just a few examples of his diverse creations.
Alexa & Stack-chan
It responds to calls in conjunction with Alexa. It reports the weather forecast.
Google translate via Siri shortcut with Stack-chan
Using the Google Translate service via Siri's shortcut feature, making Stack Chan a great translation assistant.
Such as the applications above, he makes full use of the M5Stack's WiFi communication to connect to these services.
It's also worth noting that he uses cases among Stack-chan's assets, and uses the familiar M5Stack-Avatar library and other Arduino software stacks for firmware development.
He brings out the best of Stack-chan's capabilities in many ways, but what I found most iconic about his relationship with Stack-chan was the following tweet
Isn't this also a new and ideal relationship between robots and people?
10/14/2021 at 16:10 •
Stack-chan has made it to the final round of the HackadayPrize!
Now I'm preparing for the final round, where I plan to show Stuck-chan's potential as a swarm robot. To do this, I have to keep repeating the implementation.
Today, I mounted five more boards for Stack-chan. Although I need to hand-mount small SMD components, each board takes only about 15 minutes to complete.
After using them for some demos, I plan to distribute those boards to people who do not have the environment to mount them.
10/01/2021 at 15:12 •
Smiling Stack-chan, synchronized with my face!
It can follow my face orientation, eye/mouth opening and smiling.
I used Google MediaPipe for face tracking. MediaPipe extracts facial contours and meshes with very high accuracy. I used those meshes to calculate the orientation of the face and the opening and closing of the eyes and mouth.
I used WebSocket to send information to Stack-chan. The Moddable framework includes a WebSocket module. It's easy to use and fast enough, about 20 fps in the case of the above video.
Face synchronization adds a powerful capability to Stack-chan for many use-cases such as teleexistence avatar robot.
09/26/2021 at 16:35 •
The new Stack-chan board (v0.1.1) is now compatible for M5Stack CORE2!
M5Stack CORE2 has touch screen. You can make touch interactions with Stack-chan.
Other than this the board improved many as listed below.
- Add paths between GPIO pins of COER2 to PWM servos
- you can choose Basic or CORE2 via jumper
- Add SMD diode for stability
- Add hole for mounting 25*2 pinheader
- Adjust screw hole
You can see the full difference in this pull request. And you can get PCB-manufacture-ready gerber files on my repository.
- Add paths between GPIO pins of COER2 to PWM servos
09/26/2021 at 16:15 •
The world's first ever Stack Chan workshop was held onsite!
ささみー (@sasami0709) conducted the Stack-chan workshop and photo shoot with the independently created Stack-chan features and assembly instructions.
Six participants of the workshop enjoyed learning to build robots and then went out for a photo shoot in the wilderness.
Stack-chan in the wild! So kawaii!
It should be noted that I am not directly involved in this workshop.
Since the software and hardware of Stack-chan are open to the public under the Apache 2.0 license, this kind of proprietary brush-up and even monetization is possible. I'm glad to see the circle of Stack-chan expanding, while paying people what they deserve for their work.
Good boy...or girl... or whatever...
This series of workshops will continue to be held on an irregular basis. The next one will be held in Tokyo on Saturday, October 23. Check this out!
Photos and videos by ささみー (@sasami0709)
09/23/2021 at 16:23 •
From my recent tweets(2021-08-24)
Stack-chan can speak sentences using Text to Speech function.
You can choose how Stack-han speaks, either by pre-generating a voice or by streaming a voice query on demand.
See the document for further details: https://github.com/meganetaaan/stack-chan/blob/main/firmware/docs/text-to-speech.md
09/17/2021 at 10:52 •
From recent tweets(2021-08-14)
Two Stack-chans synchronized over mDNS
One is the leader and the other is follower who mimics its pose.
src is below.
09/15/2021 at 11:30 •
From recent tweets (2021-08-07)
Stack-chan integrated with UnitV2 AI Camera by M5Stack
The camera recognizes human face and send face position to Stack-chan.
Then Stack-chan can follow the recognized face.