Introducing the OS3M Mouse, the open source 3D mouse... mouse!
HTML-код
- Опубликовано: 7 авг 2024
- Check out the hackaday page here: hackaday.io/project/187172-os...
And the different repos for building and using one here:
github.com/spoter368/os3m-har...
github.com/spoter368/os3m-fir...
github.com/spoter368/os3m-sof... 
Наука
I'd love this for Blender/game engines! One thing I wish we could have seen is footage of your hand using the device overlayed on the video of CAD navigation.
That’s a good point! I actually thought about doing that but wasn’t confident enough in my video editing skills. If you look on my channel for the previous video showing MVP of this project, there’s footage of me using it with hand in view.
They don't move much so it would be hard to really show I think. Imagine you are holding a little tiny version of what you see on screen. They way you move the sensor head is how it shifts on screen. The flip of that is Imagine the object is fixed and you are moving a camera flying around the object.
@@ColtonBaldridge how can you manage sending usb hid data from inputs and say such things about picture in picture, there's literally 40 second videos about how to do this. I'm waiting impatiently now for an updateee
I love it! I was thinking about doing something similar, but when I checked what was needed I ran into the driver problem. As far as I can tell Fusion 360 has no universal HID interface. It accepts the 3DConnexion spacemouse, but that has a proprietary driver. So even if you get it running with that driver, the manufacturer could take that whole open source project down for violating the license of the driver.
As someone who uses a 3dconnexion SpaceBall on Linux on a very regular basis, I've seen several approaches over the years people have taken to implement something similar and I can honestly say that your design is absolutely fantastic!
I love the compliant support as well. Very nicely designed, especially how you managed to keep it so compact.
Awesome job!
Edit: To clarify - this is the first one I've seen that I can actually see myself using in place of the commercial SpaceBall.
yeah, i was thinking exactly about spacemouse too.
I have a Spaceball 4000FLX. Having resurrected the old Spaceball serial driver in Linux, it shows up as a "joystick" device, and is therefore is visible in SDL2 (although it wil often try to coerce them into a "game controller"). The utility `sdl-jstest` will show you the deflection on all six axes, as well as all the buttons. If you can get your device recognized by SDL2, then you can basically use it to drive any game with joystick support.
My personal favorite is to fly the spectator camera around in Quakeworld, which still repects proportional joystick input (i.e. doesn't try to convert it to button presses).
Very nice work!
SDL2 is such a nice utility for building into a framework
Finally a real 3d mouse that's buildable.
This is it. This is the first attempt I've seen at making a spacemouse that actually implements the full 6DOF mechanism!
Keen to see where this project goes and I may see what I can do to contribute to the open source project in the future.
Amazing work :)
Awesome stuff. By far the highest potential and coolest design out of all the 3D mice projects I've seen.
Wow! Amazing work! Love the simplicity of the flexure and how minimal the BOM is with PCB coils and 5.1k standard resistors.
Amazing work! I am always blown away by projects like this. I consider myself decent with the basics of CAD, electrical/ mechanical engineering... this still just impresses me so much. Congrats on your successes so far, keep the dedication. I look forward to seeing future projects!
Best one i've seen on youtube. Seems to act like the real thing, props for your awesome design
This is the best DIY spacemouse implementation I've ever seen! Also love that you didn't stop at the HID device part, most people don't make it to the API part and try to move mouse cursor instead, resulting in only 1-axis-at-the-time movement.
Nice work Colton! This seems like a low cost solution. No magnets and no fancy magnetic chips. You can even have metal insert in the injected molded top part.
Dude the inductance method is genius.
I've been thinking about a 6dof joystick for years now.
I never could decide on what sensors to use, but this could be a real cheap and reliable solution.
3dof magnetometer are cool but they often cost to much and may be hard to get.
The real problem with 6dof joysticks is axis input separation. Still sad that Sublight Dynamics failed though.
@@DannoHung my plan was to elevate that with mechanical constraints that mimic hand and wrist motion so it's clear cut what's rotational/translational.
Currently I'm thinking of using a 2d flexure for x/y load cell for z a wrist joint for pitch/yaw and a hinge for roll.
X/Y/Z will probably be moved via some sort of cuff below the wrist and arm movement while the rotation is handled via the joystick part.
I dunno, still rough sketches nothing done in CAD yet.
@@DerSolinski Sounds interesting! Hope you come up with something great!
@@DannoHungThe biggest issue for that I haven't found a adequate solution yet is centering the entire thing...
It needs to be reliable with a linear or better a constant force.
I loathe joysticks that have weird force curves, precision tanks with those.
@@DerSolinski I'm making one with 6 hall effect sensors + 4 magnets (surrounded by thin steel sheet for magnet isolation). The structure so far is a ball joint (M6 rod end) on 3 sliding joints (3mm stainless rods + bushings). Structure needs to be mouse on ball joint on sliding joints if you don't want axes to interfere with each other. I thought a lot about using flexures but they waste a lot of space and don't have enough motion for that config.
finally someone that makes an actual space mouse rather than a fancy macropad
Wow! This is awesome, thank you for making it open source! I really admire putting all this effort and still opening it to the public
Finally! this has been needed for a while. Good job!
this is awesome! this is one of those things where i had the idea for the end product but i couldn't figure out how to get there. i'm glad someone else figured it out! my approach was to use an array of IR LEDs, but i love the three dimes and coils
IR LEDs was the approach that 3Dconnexion has used since the start, though newer models may have moved to magnetic sensors. The problem with this is that they use quite pricey sensors, one-dimensional optical position-sensitive detectors. There's also the Sixnav by ShamrockTeam/GreenCappuccino, with a build guide on electromaker that uses that uses one of the older 3Dconnexion patents, except they use a phototransistor and use varying thickness of plastic to variably block the IR output to adjust positioning. There's also the SpaceFox, which is a potentiometer-based stewart-like platform, and the Space Mushroom, which is based on a triplet of joysticks; these two, the previous, and the OS3M are the only fully-developed 6DOF projects I know of, other '3D' mice seemingly don't have all six degrees of freedom or they're underdeveloped.
I love it, more people shoud have the chance to use a spacemouse and creating open source projects around the idea helps to move towards that a lot!
Super interesting. Please keep us updated.
I think you are awesome. Figuring this out and actually making it work but then also sharing it and asking for critique? Epic mindset! Thank you for sharing it !
Awesome work!
Impressive design and commitment to it
An electromagnetic Stewart platform is an absolute genius!!!! Very nice development!!!
Amazing video, I would love to see your hand using the mouse though! Thanks for the video :)
Great work, sir!
Great project! Thank you for sharing your efforts!
Excellent work. It takes a lot to impress me, but that qualifies.
Awesome project!!
Best implementation I have seen for this. Using the Solidworks API calls for smooth, simulatneous view control is critical in making this actually useful. I have seen some other implementations that just translate into simple pan,zoom, rotate commands that I can just do with a mouse, and in some cases when I had to zoom in on details, Solidworks get finicky with those movements, and my 3D spacemouse using the API calls was the only salvation for precise viewport control.
The compliant mechanism is a fantastic idea. My best idea for a DIY one before getting a Spacemouse was using a spring loaded typical ball joint ended strut based Stewart platform.
commenting to boost the chances of this hitting the algo, amazing build !!
Fascinating and admirable, thank you for sharing!
this seems to be the only properly implemented 6 DOF solution that could really become a precision tool. Super awesome!
Very cool project! I've been looking for something cool to try making with PCBs from a supplier!
Let's go! Excited to see more of this 😁
I am so pleased to see this. I have a printer and a cad mouse like that but retail and I have a deep respect for those that toil away to make the little guy better. I hop you get it all worked out because I will buy the parts and try it. Excellent
Heck yeah! Thank you!
This is really cool!
I wonder if it's possible to emulate a real space mouse so that your version can plug-and-play with many applications
Great concept.
That looks brilliant 😀
I wish you success. This is fantastic.
Whoa! Nice!
Awesome dyi space mouse!
The inductive sensing is a cool solution. I was working on a similar device using Adafruit’s 3D magnetometer and a permanent magnet though I didn’t complete it. I had also written a Autodesk Inventor camera driver for an Xbox controller (with the plan to swap out the inputs for my DIY space mouse) but their camera motion is really weird and jittery, not like the real Spacemouse.
Incredible💪👏
Amazing! I built a Orbion mouse and was unfortunately pretty disappointed. I tried it twice, set it aside, and haven’t used it since. I think this would be a much better solution!
Finally, an OS3M mouse which is an HID device that supports 6 DOF freedom for CAD design. I'm on my way to the ATM machine to withdraw some USD dollars to purchase this right now
A bold project. Love it.
Great work :D
sick af
I bought a commercial space mouse for about ten years ago and this is my reaction to this project: Awesome!
I'd love to collab on making a machined metal one with ya, would be insane!
It's great that someone is making an actual open-source 3d mouse analog, and not just a joystick that moves mouse from the center of the viewport to the sides.
Though I want to be optimistic, but it seems you project is very close to the way the actual 3dConnexion products work.
They may file for patent violation.
But I also think, the most important part here would be actual software implementation, since there is a lot of way to input 6d0f that aren't patented by 3dConnexion.
3DConnexion uses some LEDs and linear optical resistors to detect where the light hits the resistor with some slits. So the electronics are quite different.
@@PampersRockaerOh, yeah. I heard that they used something optical, but never found the good teardown.
@@trashviewer3521 If you are interested, look up "What's Inside a Space Mouse?" from
"Thought Bomb Design".
Amazing!
I want to lead with saying I barley understand any of the technical information you spoke about in this video. With that out of the way, Id like to share a story.
For years, I have been saying that computer mice need to be turned into "next gen" by using 2 optical sensors to precisely calculate rotation. I mainly figured this for gaming purposes. Helicopter controls in battlefield 4 was my real pipe dream. Everyone I told loved my idea but I have no idea how to make something like that.
This reminds me of that idea. I hope you find nothing but immense success from this. Would love to see you mess around with as many different uses for this as you can.
Definitely subscribing.
The main problem with two optical sensors is that your hand already rotates a bit while using a mouse.
@@totally_not_a_bot I feel like it would be a nice "optional" axis, where most applications will only care about translation but the rotation is available for when you want an application to pay attention to it. Kind of like how you can map game controllers to behave as a mouse with the thumb stick even though they have so many other buttons in actual games. If something is designed with it in mind then it wouldn't be as much of an issue, and if not it's not in the way of anything, just unused data
@@DakkyW Take your mouse, disconnect it, twist it 15-30 degrees while gripping normally and move it around some. It's uncomfortable in either direction, moreso to the outside. It could maybe work with a momentary switch under the thumb, but the idea of twisting my wrist like that seems painful.
@@totally_not_a_bot I have a good idea then for you. When something like that comes out, don't buy it! Tired of negative nancys on the internet. I have my mouse in my hand right now and using my thumb and ring finger as points of contacts, I can easily rotate the mouse almost a full 180 degrees. And like the other guy said, use it like a regular mouse the other 90% of the time.
Amazing, finally a real open source 3D mouse!
Cool i was thinking of having this as a project myself.
Very cool!
the inductive sensing concept is interesting. I wonder if you could get better linearity with something like an LVDT where there are two opposed inductors in series and they cancel out in the zero position. apart from CAD it'd be cool to use one of these to do RCS manoeuvres in KSP.
Hadn't heard of LVDT's until now. You'd have to wind your own and have six 5khz Sine generators. The idea is very possible, but maybe too bulky with DIY methods.
@@ColinJWiens You can make one on a PCB, they're not that difficult, it's just more flat coils. It's been awhile since I've looked, but I believe quite a lot of inductive true position sensors, especially rotary ones, do use a variable transmission rather than the typical basic inductive sensor setup, where a lot of modern sensors are PCB-based and not the typical wound coil construction. The only reasonable way to make this work is to have power through the knob and have the upper be the energized side, with the sensing coils on the base, though doing this would also remove the need for a target as the primary would act as the target.
I also saw your comment in a different thread about using hall effect sensors. Do yourself a favor and use MR sensors instead, they're basically a wheatstone bridge of magnetoresistive elements, they're a lot less noisy and have much better centers, plus hall sensors have other downsides whereas the MR sensor's only true downside is that it's suceptible to external fields though I haven't yet seen one that's properly shielded.
This looks awesome, I might make one myself in the future!
Have you considered adding some buttons on the handle?
I did actually think about that! Buttons on the knob itself would be pretty hard to do, as the flexure gets in the way and you’d have a hard time getting wires to them. The alternative I’ve provided is an I2C header on the PCB, for other expansion PCBs that have yet to be designed. I’m not sure if I’ll ever design them or add them to the firmware as it’s not really something useful to me, but since it’s open source, someone else totally could!
@@ColtonBaldridge maybe you could have an inductor coil in the middle that runs to different resistors based on which button(s) you press, and then you use a stationary inductive coil at the base to pick up the change in frequency or phase (I have no idea if phase is affected or not), and then correct for the estimated distance by using the kinematics collected from the other inductors
would love to see a metric version of this!
Awesome!
awesome :) i have a 3dconnexion spaceball 5000 but its sooooo big
Wow! Yes, would be cool to buy a PCB.
Amazing! I use my Quest 2 controllers for 3D in Blender, but back in the day I wanted to make this so much. I used a laser shining into a ball of hot glue to make a super bright glowing point, then tracked that with a camera, but I didn't understand AI well enough to get the tracker working reliably. :-/
Crazy cool solution lol
What do I think? I think you're pretty darn clever. Good applied physics, geometry, software, and fine electron pushing.
would be interesting to add a place to put the edge of your hand and an optical sensor at the bottom and use it as a cursor mouse aswell. to free up your left hand for keyboard shortcuts etc
finally an actual 3d mouse with true 6dof
3dconnexion seriously needs alternatives if we want them to lower their prices.
Also using inductance instead of LEDs with phototransistors feels like a good path, it wont be sensitive to dust or other ingress unlike the 3dconnexion mice that require a dust shield between the moving part and the base. Also your approach eliminates the need to route power to the joystick handle, which is really good. On the spacemouse they do that through the springs that are soldered between the two PCBs. It's somewhat elegant but it makes assembly and disassembly really annoying
Been thinking of making a hall effect sensor version of this... this is much more elegant!! As peeved as I am at how much better this is. this is such a great idea! will definitely try this one at home. I might remix this so that it can also function as a joystick by having replaceable knobs. also thinking of adding an extra lot of buttons for shortcuts (similar to the 3dconexion mouse!). let me know know if you are interested in my remixes :)
Liked and subbed.
Fantastic! A design always 'clicks' with me when it's simultaneously so simple (in BOM terms) and complex at the same time. I'm seriously considering adding this to my build queue. Will you be doing a small run of PCBs/3d parts for sale?
Dude, I bought a resin printer. I'm still learning to cope with setting that shit up, since no print would stick to the plate. But as soon as I solve that issue, I'm so stealing this.
It looks like a great first project to mix some electronics with 3d printing that's actually interestingly functional.
impressive
Amazing work, did it works on autodesk inventor?
Would love to build one, is there somewhere I can buy the PCB from you?
nice
Neato
Very nice, I'd like to see how the device itself is used too though lol
If you rig up a wii-style 3-axis gyro you could make something like a 6-dimensional mpise ball (all rotations and translations) with 100% reliability which would be great for modelling
Man, looks nice, two questions:
1. When you are planning release ?
2. Maybe something else, but not dimes ? I'm from completely different country, but still would love to try build this.
Already tried different diy project of such a device abd it's not good at all, but your one seems better.
I’m not sure what you mean by release? You could build it right now if you wanted to. All the files are in the github repos
And you can use anything aluminum or copper (conductive metals) in place of the dimes.
Cool
Really cool. The use of the coil seems very innovative. I did not heard of anyone implementing a HID device in this particular way.
Do you think this approach is better than using a 3d magnetometer like a GY50 or an adafruit MMA8451?
Yeah, I do believe this approach is better. I think capturing translational motion with an accelerometer/magnetometer would be really difficult, but when I was first thinking about this problem it is something I considered! Thanks for watching!
@@ColtonBaldridge I think that 3dconnexion have patented using hall effects, so this approach is probably also ideal legally
@@HorizonMakes Link it if you find it but I briefly searched their patents and didn't see the word "hall" anywhere nor a mention of magnet-based sensors. You might be thinking of their "optoelectronic" sensor patent. I am not a lawyer.
Cool!! 👍🤓
Very nice build for someone who nearly understands it himself! 😅
I see 6 coils on the PCB is that correct!?
And 3 metal coins.
So I guess it works a bit like a hall effect sensor array!? 🤔
I wonder what would happen if you would replace the coins with neodymium magnets,
Would it increase the accuracy?! 🤷🏻♂️🤔
You may want an hall effect sensor if you're using magnets, the nature of these inductive sensor coils is that they are what's creating the magnetic field and sensing how the current behaves in doing so. Using a magnet would create a static magnetic field which wouldn't really effect the power used to create the induced field as a non-moving magnetic field doesn't induce current. The metal in the magnet itself would still have an effect the same as any similarly ferrous metal, magnetic or not.
nice ..
Teaching Tech dud one that emulates a 3DSpace mouse, could you use it to make this mouse emulate it too?
Looks cool. Could it be use like hotas?
Yup, I’ve tested it out in Elite: Dangerous and it works with no additional software required
I bought a really nice 3D Mouse about a decade ago... and never found a good use for it. It was super cool, but for 3D apps, you really only want a little movement that is really refined, like a single axis mouse drag. The 3D Mouse did too much and yet too little. I found it frustrating to use. People have been trying to replace the humble mouse for decades, but it's just so darned accurate and stable.
All depends on the application. 2D mouse will always be king for selecting/aiming in 2D. Even for 3D, a 3D mouse may or may not be best depending on the application. For many games, using 4 axes from a 3D joystick (which isn't commercially available) in combination with the 2D mouse may be best - based on a review I watched of a failed 6Dof kickstarter joystick.
This is awesome! I'm wondering, what's the feel like? Does it feel fairly sturdy? Or more like it's floating in air?
Thanks! I’d say it’s somewhere in the middle. There’s a decent amount of resistance, where I’ve built it to be symmetrical about all axes. It returns to center pretty quickly when you let go of it, but the response is a bit undamped, so sometimes it’ll shake a bit when you let go.
@@ColtonBaldridge that's so cool! Have you tried different filaments to feel the differences between them? Which are you using?
@@nadimfetaih3841 I only have PLA, so that’s what I’ve been designing around and using. I’d love for someone else to use the files and print it out of something else though!
between this and the beacon abl, i really need an excuse to use one of those LDCs in a project
They’re amazing little devices, that’s for sure. They also have some weird clocking quirks and the datasheet does not do a great job of explaining everything, so be wary in that regard. Either way, I haven’t been able to find anything else out there like them, so it’s hard to complain.
What can we use instead of dimes ? Currency around the world uses different alloys. Is it just a matter of calibration?
Any conductive material can be used. Aluminum and copper are some of the best options. And yes, you’d have to recalibrate a few constants in the code for other targets.
Does this work in fusion 360?
Very nice work there! Are you using the plastic from the knob to act as springs? Any concerns about long term strength of that?
There is a separate flexure I’ve designed to be 3D printed inside the knob, which I think is cool because you can change it however you want (more or less stiff in any axis). Strength wise, it’s been sufficient. I have broken them, but only when applying loads I’d consider to be excessive. Even if you do break them though, another one is only a 10m 3D print away; you don’t have to print a whole extra knob.
why not a stabilized gyroscope ?
could this be done with hall effect sensors
yeah, assuming you make sure the nearby magnets don't mess with unrelated hall effect sensors in an undesired way. Could be done with 3 magnets + 6 sensors or 2 magnets + 6 sensors (assuming the 2 magnets aren't in the same axis)
ahaha I thought this was gonna be a completely different kind of project when I saw the title on my home page. I assume that confusion is exactly why this is called OS3M and not OSSM. 😉
Btw cool space mouse. Induction really is the best way to tackle these sort of mice. Ive tried making ones that use multiple joysticks in the past and they just never tracked very well.
Would be cool if you could have some form mechanical lockout switches in the flexure, to allow you to bind one or more axis' so only the remaining axis' can be flexed and interpreted. Could also be done in software but mechanically binding the axis' could allow the buttons to be on the knob itself. Not sure if its possible tho just an idea.
why didn’t you use springs?
Hi there, is there a comprehensive tutorial or instruction to build this mouse. I´m not really deep in all this and would appreciate it.
Hi! Not currently, as the project is still very much in development. As things solidify more, I’ll be making one though, and possibly even selling complete PCBs!
Can you please make a video about the flexure design, or link to some info on its design process?
Truthfully, the flexure was designed purely iteratively. That is, I kept trying designs until I got to one I felt had resistance balanced along all 6 axes (or close enough, at least). My background is EE, so the mechanical stuff is out of my wheelhouse. I plan on making another video going more in depth into the design and how to build it soon(ish) after I finish some upgrades.
Oh, I will say you should checkout the hackaday link in the description though. One of the logs I wrote was dedicated to the flexure and solidworks simulations I did on it (that ended up not being realistic or worthwhile).
Cool Ill check it out...thanks for the reply!@@ColtonBaldridge
What is used as a target?
There doesn't seem to be a BoM telling and they are a bit washed out in the video.
Looks like some kind of national currency that probably contains much copper?
Yeah, apologies, the HDR to SDR conversion wiped out a lot of the clarity in the video. At the beginning I mention the BoM is the 3D printed parts, the PCB, 9 M3x10 fasteners, and 3 dimes (USA 10 cent coin). But as you mention, any object of similar size with copper inside will be fine. I’m working on a follow-up video that goes into more clarity on how to build one. Thanks for watching!
It seems like you are getting some love from the comments
Will you be making those? I don't have necessary tools, materials, or skills to make one, even with schematics, but would pay for one if it came prebuilt.
Not currently, but maybe in the future!