As someone who finished their undergrad it goes to further my self doubts as I loved matlab in college and his 3d implementation makes anything I’ve ever done seem trivial, and I don’t even know how one would begin FEA in matlab. I assume plug-ins of some kind.
as an undergraduate pursueing ME in my 3rd year this is the kinda stuff that makes me feel even more useless when the workload gets crazy because i could never come close to writing the matlab scrips for this
@@daneprostamo-brown6552 I loved matlab, so this did a similar thing, but remember that the purpose of masters subjects is to delve very deep into your research project. This could have been a year in the making.
@@daneprostamo-brown6552 thanks for this comment, for a bit I felt as if having smol brein exactly because of math involved. apparently this sensation is quite common here :D
Really well written, performed and edited; a brilliant showcase of how video summaries of journal articles and wider academic research should be done. If I had a hat, I would have taken it off.
I am but a simple man that has always been fascinated at machines/devices that have reached their performance limit. This seems to almost be "nature like efficient", just brilliant. But most of all, it was explained to me in a way that I am very grateful for. Thank You.
I almost want to take this flexblade concept and make it work with magnetic sensors rather than voice coils. I've always been interesting in 6dof platforms for use as input devices, as more things become fully 3D we also need proper 3D inputs, and those inputs will more than likely be better as a combined manipulator; there's a reason why we use mice and joysticks for 2D manipulation, rather than separated single-axis manipulators in unison for planar position manipulation. Currently I believe the 3Dconnexion SpaceMouse probably has the best method for a low-throw peripheral, especially for manufacturability, though I personally want something with a larger throw and higher resolution along its movement; the only problem with these mice is that there's next to no information on construction or the method of how they're achieving six degrees of freedom, I had to rely on an obscure blog post to find one torn down, and I'm not sure how fruitful attempting to find patents would be. I've toyed around with Stewart platforms in the past, but there's just too many downfalls when reversing the mechanism for use as an input, servo-based ones aren't optimal, high resolution encoders gets costly, moving to a linear design is improbable as linear encoders aren't as common and the translation from linear to rotational movement includes losses, etc. This flexblade concept may prove to be a better alternative, magnetic sensors are, by comparison, cheap and fairly reliable, and the compliant mechanism itself is fairly simple in comparison, and could potentially be made into a volumetrically smaller package with some modification; alternatively, other depth-based sensors could be used for this, magnets are just the simplest option that came to mind, and more than likely the cheapest at the end of the day. Not only this, but this kind of design could also be translated to a mechanism that is only 3dof if a simpler input scheme is needed. The only thing is, with the modified design to make this an input device rather than an output device, how far can you actually push the throw of the platform, or if the design is inherently limited to a short throw as the original design was intended for micron levels of resolution for a printer head.
I watched a series of videos on YT where a guy made a controller similar to 3dconnexion mouse, all by himself, an advanced DIY. I'll try to find it and link it here, if not try googling it. Anyway, I really like your idea of 3d controller, I hope you will get there and enter the market with revolutionary input device and make lots of money. Oh, just now i got an idea, with all current advancements in AI, computer vision and whatnot, maybe it will be feasible to just use an empty box with 3(?) cameras that will track hand movements and translate these to inputs. Hand could be in a glove or maybe some rings or sth like that to better discern between fingers that can have different functions? Not sure if something like this isn't already in development, if I'm first and someone actually make such device please throw me some $$$ when you commercialize my idea :)
@@aeonikus1 I've seen multiple projects for a 6DOF joystick like the SpaceMouse, though the only video series that comes to mind is Space Mushroom by @shiura, which uses three dual-axis joysticks and a slider mechanism for various freedoms and constraints. Is this the one you're speaking of? I'd also be careful with using the term '3D' when talking about joysticks, it's such a loose term that doesn't mean much, and there's reason for that. There's 2-axis joysticks like what we see on nearly everything; and they can be considered '3D' for a simple reason, imagine you're at the center of a sphere, you can locate any coordinate without having a degree of freedom, rotation in this case, within the axis of depth, and this is quite literally how a camera stick within a video game works, and inversely you can navigate a sphere in two axes without additional degrees of freedom based on how a planar grid can loosely map to a sphere, which is how a movement stick within a video game works though this applies to really any kind of movement grid not just planar or spherical. Then there's 3-axis joysticks, which add a third axis to the stick in some way, typically it's just a Z-rotation under the Xr-Yr gimbal, but these see very limited uses outside of some niche applications, basically anywhere where 3DOF manipulation is needed. Then we get into 4DOF joysticks, which are typically some kind of X-Y manipulator on top of another X-Y manipulator, such as a slider or an analog pointing stick on a typical joystick, though at this point the joystick itself becomes a full-sized hand joystick due to obvious dexterity issues, where the new upper manipulator becomes the thumb's joystick. 5DOF, though rare if they even do exist, would be a combination of the two previous, a 4DOF on top of a Z rotation. Then we get into 6DOF, which are entirely different in that they read both linear and rotated movements of the manipulator in a more complex sensor setup, again due to dexterity limitations; I guess you could take the 5DOF and put a secondary Z rotation under your finger as a bi-directional form of trigger. All of these of which can be considered '3D' because of how we define what 3D is, and that's typically not in clean-cut dimensional axes but rather how we manipulate objects, of which objects are typically 3D, so it's better to define a manipulator by the degrees of freedom it has in total and further refine that by how it achieves those degrees of freedom; for example, the modern gamepad is 5DOF, but it uses two 2-axis joysticks and two triggers on a shared axis. Triggers themselves are typically only a half axis, they can be a full-axis but they're almost never logically implemented as such, probably due to the fact they only move in one direction rather than two; which is why I previously mentioned a bi-directional 'trigger,' which wouldn't be difficult to produce, it's just that nobody really makes them. There's also meaning in the term of 'joystick' modernly as an analog stick, joysticks used to be digital and there's very little implementation of such outside of arcade-like uses, but even then a joystick has to have a return to center mechanism, otherwise it's a continuous stepper which is why dial-like devices should be excluded when talking about a degree of freedom, as a DOF should be free to be manipulated rather than detented to specific steps; of which people that don't know better will probably get confused and just suggest that glorified buttons in the form of a stepped dial should be a 'solution' to adding a degree of freedom, sure our scroll wheel on our mice might work for zooming but we're also actuating that in steps rather in a continual input. At least this is true for manipulated inputs, the term 'degree of freedom' is also dependant on use; and some might argue that digital inputs can be a degree of freedom, but I don't really consider pressing a key of stepped input equal to the amount of what an analog sensor outputs is really a 'degree of freedom,' and I know someone will be pedantic enough to argue that 'it's all digital anyways, because that's how USB communicates, and analog devices use an ADC, etc.,' but that's really beyond this argument considering a stepped input is still one-bit binary whereas an ADC will typically be 8-bit if not 12 or 16, so even by that argument anlaog, whicle digital in the end, is still much more refined. Mice (trackballs included) and trackpads are in a limbo between the two, as they're essentially a vector device, but they're still not 'continuous' in the sense of what an analog sensor reads as, not to mention their physical, and often digital, boundaries. Input devices are more complicated than just calling something a '3D joystick,' it's better to use more refined terms, and to define how the input device actually reads its input.
@@TankR No, it's not. Waldos are also for digital puppeteering, not animatronics. Not to mention how large these devices are when, if you had any amount of reading comprehension, the discussed topic is about mouse-sized devices. But given your entirely incorrect use, and incorrect formatting, ellipses, I doubt you're literate enough to understand literally anything within this comment thread. Here's the main issue, Waldo is just a name for a range of devices, and a Waldo is purpose-built for a specific rig in digital puppetry, whether that puppet itself is digital or physical; again, none of this is on the animatronics side of a physical puppet. A great example of this is the difference between Mr. Volt's (@MrVolt) Wheatley puppet's Waldo and production Waldos such as what Henson studios is doing for their Waldo rigs; the former is purpose built for the various degrees of freedom a specific puppet needs, the latter is more open to multiple puppets but is widely used as only a head rig while alternative puppeteering rigs control the body of the puppets- and here's the thing, the vast majority of Waldos are discretely for head and neck manipulators. Whereas what is being discussed here, stemming from what the original device in the video was as I discussed literally only changing actuator to sensor, i.e. to make a driven output device into a driven input device, is a 6dof manipulator that's small enough in size to be used by a hand on a desktop; I've even mentioned the SpaceMouse and its DIYed alternatives, this should've been your second clue, right after watching the video and presumably having an understanding of the original device. And here's the next thing, how degrees of freedom are used in naming, what the SpaceMouse is, its alternatives, and what I suggested in my original comment are three linear axes and three rotational axes on the same manipulated part, whereas Waldos on the other hand are rarely considered '6dof' in this way, nor are executed as such, very similar to how robot degrees of freedom are otherwise named and calculated; you're comparing apples to oranges, hell at this point it might as well be une comparaison de pommes aux pommes de terre, if you understand French and have enough logical understanding of why it's a bad comparison via naming convention, of which I'm doubtful of you having any capabilities of logical thought, comprehension competency, or even basic understanding period. The next issue is size, Waldos are absolutely massive and complex. You might as well recreate a Novint Falcon for a 6dof manipulator at this point if you're arguing a Waldo would be of any practical use here. Again, your first clue should've been the fact that the device in the video is driven by voice coils, and you presumably understand how utterly small voice coils can get, given that a lot of IEMs (earbuds) are dynamic drivers, i.e. driven by voice coils. Again, your second clue should've been the relevant devices that have been named dropped throughout this comment thread, even if you don't know the device itself it is otherwise easily searchable; and stemming from this, even if you didn't have relevant knowledge of the named devices, you have analogous knowledge considering they're compared to a computer mouse. And no, Waldos cannot be reasonably sized down, considering they encapsulate your hand and wrist at minimum, then on top of that you also have a stable frame to keep them in place or you have a perch that sits under your forearm. Even if you could size down a Waldo to even properly fit on a desk and integrate into a peripheral ecosystem, it would again akin to be what the size of the Novint Falcon was, if not even large; and again here, I'm thinking you lack any amount of logical thought, comprehension competency, and basic understanding of things. So, what does this mean? Waldos are nonapplicable for the argument and are irrelevant to the conversation, and they cannot be made to work in a way that makes them applicable to the conversation. You lack any amount of reading comprehension competency, literacy, logical thought, the capability to research anything, and lack even the basic understanding of anything relevant to the conversation nor the basic understanding of what you're even talking about. Perhaps you should educate yourself, and until the point of being educated I would suggest that you sit down and shut up. Your reply is on-par with some of the most misguided, misinformed, and unintelligent horseshit I've read in relation to intelligent conversation. Even out of respect for intelligent conversation, and by extension respect for others' time and intelligence, you could have at least done your due diligence to understand the conversation itself, rather than spewing nonsense that is proven to be irrelevant, especially when you do not inherently understand what you yourself are talking about. You can't even have respect for basic conversation, any intelligible portions of the conversation aside, as your incorrect use of ellipses heavily implies a tone that I, or by extension the other party as well, have no idea of what we're talking about and implies a tone that you are talking down to us as if we're lesser, pauses in speech as how you executed always imply this tone, in verbal speech, in writing, and in literacy uses such as dialog in story writing; and this is a point you can't weasel your way out of. People like you act like they're so intelligent, yet are some of the most unintelligent, act as if you're so highly informed, yet are rather the least informed, then you go off on this twatty behavior such as the tonal implications of how you formatted your reply, yet you're the one that's entirely deserving of being talked down to; you should be embarrassed of yourself, I'm sure others are, so again you should sit down, shut up, educate yourself, learn some very basic life skills such as logical thought and comprehension competency, and learn some basic respect for literally anything you interact with. Given that your account has existed since 2013, that implies that you're easily in your 20s if not older, things like this should not have to be explained to you, I get it if the education system and your parents failed you as a child, but you've failed yourself in your adult life, you were likely a class clown but now you're outright just a clown that provides nothing but dogshit on a silver platter. And on an academic channel, where most people are active in intelligent conversation, no less. And on a comment that's actually speaking of the mechanism itself and having advancements elsewhere, no less. Do better, a cat pissing in sand is more sophisticated than you'll ever be.
Seriously awesome work. Bridging mathematical, computational, and real world testing is a lot of hard work. Mathematical analysis was awesome, FEA was excellent, and real world validation was great to see. Lots of varying feedback, in the comments, but for what this is, I think your team shot a bullseye!
I love this kind of format, a quick overview of a journal paper. Especially useful for mechatronic systems as it is much easier to understand movement of systems in animations rather than static images. As for the content, In particular I like that you kind of explored/mapped out the entire design space (7:15). That must have been quite computationally intensive, did it take a long time? 0:55 Having seen the T-flex stage in real life is very cool. Not just the folded leaf spring spherical joints are very cool, there's lots of little tricks applied in the actuation as well. Also, I noticed that the stage/platform of the hexblade is constructed out of a polymer, is this done on purpose to have some compliance in the platform? Or perhaps to make it quick and easy to exchange for other types of end stages?
How much would it cost to use this technology to build a small robot/vibrator for the female market? I'm pretty sure we could make some money with something like this. We could also train a neural network using electrodes on a few girls to find the best algorithms.
This is fucking awesome. I don't care about any applications of the tech even. Your analysis and validation of the mechanical behavior is impressive enough.
Whoo hooooo voice coils rock! :-) If you need to stabilize that which cannot be seen with the naked eye, always a great option - love the flexure action of soft robotics. The Future is NOW. Thanks for the video!
Everything was created in beautiful and most effective way. Why! Because the creator is The Affectionate and have the knowledge of all things. He gave us physics to show us how hard and beautiful his creation is, So we love him for giving us all of it.
Great video! I took your class last year Hopkins and this was a great refresher on what I've learned from then. Good job, I like the music choice and the script was interesting and easy to follow (if you are an ME i suppose) :)
This was so fascinating. I especially appreciate the use of voice coils as actuators. Were there development problems or scope issues that you encountered at any point along the process that created more work that you expected?
I love the design and the use of the matlab script you used to calculate it. My one complaint is that 7075 is poor in fatigue, and this seems like a design ripe for a fatigue failure after a few thousand hours of operation. For such thin flexures there are many steel alloys that would be superior at a minor weight penalty.
A print head mounted on one of these platforms could print a line path in non-flat layers that are optimized for physical properties like resonance and flexure. It might be useful to be able to directly generate the Gcode to FDM print a (fibre reinforced?) PLA/nylon/PC positive for this platform's frame from. Perhaps simulate the FDM line structure as well as the shape. So, even more maths. If implemented in an online calculator it would need a very long progress bar and probably a PayPal button.
Very well done. Every design choice was justified quantitatively and rigorously. The fact that the experimental natural frequency corresponded so well with your predictions is very impressive! What sort of resolution and bandwidth can you achieve with that camera-based sensing system?
Great video, awesome work. Still baffled by the sampling rate of the camera array. I would have used a 6DOF piezo sensor, because 6 kS/s for 120hz would be nice to have a good amount of datapoints to look out for harmonics, and it would have been way cheaper.
Okay, first off I see a massive amount of work went into this presentation. Secondly I love Solid works I don't think that there is a better design tool available. As to my point: How much weight is acceptable for the printer head or are you constrained to a mirror?
I'd be interested in seeing the FEA stress analysis and fatigue life of the parts made from 7075 Aluminium. I can visually see the displacement so that would be more valuable information than URES.
Fantastic design and demonstration. I just wondered why the coil isn't fixed and the external cup moves so as to remove any flexure to the driving wires? That would remove any chance of fatigue that may cause failure, however unlikely given the small range of movement? Thanks for the brilliant demo.
Nice! Although I have to wonder, once physically built, how lengthy was the calibration procedure? I'd imagine that any inexact placement of the coils would result in non-orthogonal motion?
AYYY Yo that DARPA research is MAD. Got so many ideas I want to test now. Im in my last year of my ME degree but I plan on doing research during summer involving flexure devices for sub-micron precision with vibration.
I’m curious about the folded flexures which act not just as simple bending elements, but also experience significant torsional deformation in certain movements. At first glance this would seem to make it hard to predict the position of the stage since the connection to the voice coil has multiple degrees of compliance, and deform in different ways depending on what the other voice coils are doing. I guess I’m wondering if there is “crosstalk” between the axes.
Given their thin rectangular cross section, each flexure is far more stiff in the direction of intended motion, and bends very easily in the case of not. If you watch the animations at the start, I hope that makes sense. You're likely talking when motion is not perfectly defined by 3 actuators at once, but I assumed the bending moment would still mostly be in line with the flexure's intended direction of bending, so it wouldn't be too disruptive (since its cross section is like 20 times as thick in one direction than the other, and it only receives a small portion of the load that's off axis in the first place. Hope that makes sense?
@@jasondean37 I get it, but yes, I am talking about the twisting induced by the movement of the other axes. The prototype appears to be open loop, and I think it would be quite hard to create an algorithm which would tell you what voltage the voice coils should have for a given position output. If the voice coils had encoders it would be nearly as difficult. Maybe if the end effector could be directly measured for position feedback it could work.
its interesting, im not trying to knock your work or anything, but it seems like the amount of movement it has, is so tiny, especially considereing the size, amounts of parts, and complexity it is to make. Would it scale to have much greater movement somehow?
The problem the design was meant to solve is explicitly a very small fast & precise area of movement relative to the mechanism. You are asking to solve a different problem.
Sorry I can't give it out for free or I will be hit with a copyright infringement :( If you have access to a friend at a university or company that pays journal fees though you can get it for free from them. Good luck.
No, there are six independent actuators for the six DOFs. To understand redundancy in actuation and constraint, you can watch the first and last lectures of my compliant mechanism lecture serries.
@@TheFACTsofMechanicalDesign thank you so much for putting those up. I'm only on lecture 7. So far on the FACT complete library of parallel systems I've only found 5 degrees of freedom so I haven't gotten there yet/my brain is too tiny to even understand why 5 degrees of freedom would be necessary. But in 3d printers they only use one z stepper which is so interesting because on this project it seems like the z axis is in part the mechanism crawling on the part! So I expect that the printer has to incrementally print in the z between crawl steps which is also so beautiful. In deltas, it feels like some of the goal is to reduce the number and cost of actuators and modern printers only use 3. Of course they've got a lot of extra mass and don't get to play with linear motors this excluding the need for linear actuators and 3d tracking for to the cost. I guess use of more degrees if freedom must have something to do with decoupling the different movements and it must have more capabilities than a parallel linkage in a typical printer. Anyways, thank you so much for your response and I can't wait to finish the rest of the lectures. I'm sure it will all make sense after I finish. Thank you again so much for putting them up!
It would be interesting to remove the voice coil and instead use toothed magnets (stepper-style) along the non-grounded (currently voice coil) bump stop and then coils on the grounded frame for feedback sensing and a more constrained initiation/execution of movement.
Meanwhile, down on Earth for personal purposes, a 3 degree of freedom flexure-based system that could be easily 3D-printed on typical low-end printers would be wonderful for reducing parts count and (hopefully) wear issues. Well, one can dream, right?
"large range of motion" (looks like it goes 2 inches, maybe) Seems like a better solution would just be to have 2 rigid bearing mounts to straddle the beam as it is being built, the mounts themselves could be adjusted in very small increments within the rigidity in order to maintain precision. A laser guide would be optimal along the path....and the whole thing would cost far less and take up less space.
Very cool concept and video :) Insane stuff. One critique: while it's probably fine for the intended audience (who else would click on a video with this title?), the language was very technical and jargon-y. I think this video could have been improved with some less mentally demanding writing.
"flexture ... gobldegook ... parameters ... gobldegook ...." etc. Didn't understand a word of it Which part is the flexture? What part is the parameters? Which part is the gobldegook? Not a very edumakational video
Awesome video. I really like the analytical optimization to get the pareto frontier. Have you ever done that in higher dimensions?
very interesting observation.
Shuu!!
Get out of here that knowledge will only be used to annoy your wife in one form or another...
Greatly to my entertainment that is 🤣
hi shane :)
My dude! I have no idea what any of that means but now I'll get stuck in a Google rabbit hole figuring it out, thanks.
ah
so this is why youtube recommended me this video
As an undergrad pursuing ME, this is the kind of stuff that keeps me interested when the workload gets crazy. Thank you.
As someone who finished their undergrad it goes to further my self doubts as I loved matlab in college and his 3d implementation makes anything I’ve ever done seem trivial, and I don’t even know how one would begin FEA in matlab. I assume plug-ins of some kind.
Oh I’m an idiot. I was only half watching on my first watch through and now noticed he simply used solidworks for the FEA
as an undergraduate pursueing ME in my 3rd year this is the kinda stuff that makes me feel even more useless when the workload gets crazy because i could never come close to writing the matlab scrips for this
@@daneprostamo-brown6552 I loved matlab, so this did a similar thing, but remember that the purpose of masters subjects is to delve very deep into your research project. This could have been a year in the making.
@@daneprostamo-brown6552 thanks for this comment, for a bit I felt as if having smol brein exactly because of math involved. apparently this sensation is quite common here :D
Congrats to Z. Yang for the achievement, and kudos to the prof for so clearly acknowledging the students who do much of the grunt work.
Really well written, performed and edited; a brilliant showcase of how video summaries of journal articles and wider academic research should be done. If I had a hat, I would have taken it off.
I am but a simple man that has always been fascinated at machines/devices that have reached their performance limit. This seems to almost be "nature like efficient", just brilliant. But most of all, it was explained to me in a way that I am very grateful for. Thank You.
I almost want to take this flexblade concept and make it work with magnetic sensors rather than voice coils. I've always been interesting in 6dof platforms for use as input devices, as more things become fully 3D we also need proper 3D inputs, and those inputs will more than likely be better as a combined manipulator; there's a reason why we use mice and joysticks for 2D manipulation, rather than separated single-axis manipulators in unison for planar position manipulation. Currently I believe the 3Dconnexion SpaceMouse probably has the best method for a low-throw peripheral, especially for manufacturability, though I personally want something with a larger throw and higher resolution along its movement; the only problem with these mice is that there's next to no information on construction or the method of how they're achieving six degrees of freedom, I had to rely on an obscure blog post to find one torn down, and I'm not sure how fruitful attempting to find patents would be. I've toyed around with Stewart platforms in the past, but there's just too many downfalls when reversing the mechanism for use as an input, servo-based ones aren't optimal, high resolution encoders gets costly, moving to a linear design is improbable as linear encoders aren't as common and the translation from linear to rotational movement includes losses, etc. This flexblade concept may prove to be a better alternative, magnetic sensors are, by comparison, cheap and fairly reliable, and the compliant mechanism itself is fairly simple in comparison, and could potentially be made into a volumetrically smaller package with some modification; alternatively, other depth-based sensors could be used for this, magnets are just the simplest option that came to mind, and more than likely the cheapest at the end of the day. Not only this, but this kind of design could also be translated to a mechanism that is only 3dof if a simpler input scheme is needed. The only thing is, with the modified design to make this an input device rather than an output device, how far can you actually push the throw of the platform, or if the design is inherently limited to a short throw as the original design was intended for micron levels of resolution for a printer head.
I watched a series of videos on YT where a guy made a controller similar to 3dconnexion mouse, all by himself, an advanced DIY. I'll try to find it and link it here, if not try googling it. Anyway, I really like your idea of 3d controller, I hope you will get there and enter the market with revolutionary input device and make lots of money.
Oh, just now i got an idea, with all current advancements in AI, computer vision and whatnot, maybe it will be feasible to just use an empty box with 3(?) cameras that will track hand movements and translate these to inputs. Hand could be in a glove or maybe some rings or sth like that to better discern between fingers that can have different functions? Not sure if something like this isn't already in development, if I'm first and someone actually make such device please throw me some $$$ when you commercialize my idea :)
@@aeonikus1 I've seen multiple projects for a 6DOF joystick like the SpaceMouse, though the only video series that comes to mind is Space Mushroom by @shiura, which uses three dual-axis joysticks and a slider mechanism for various freedoms and constraints. Is this the one you're speaking of?
I'd also be careful with using the term '3D' when talking about joysticks, it's such a loose term that doesn't mean much, and there's reason for that. There's 2-axis joysticks like what we see on nearly everything; and they can be considered '3D' for a simple reason, imagine you're at the center of a sphere, you can locate any coordinate without having a degree of freedom, rotation in this case, within the axis of depth, and this is quite literally how a camera stick within a video game works, and inversely you can navigate a sphere in two axes without additional degrees of freedom based on how a planar grid can loosely map to a sphere, which is how a movement stick within a video game works though this applies to really any kind of movement grid not just planar or spherical. Then there's 3-axis joysticks, which add a third axis to the stick in some way, typically it's just a Z-rotation under the Xr-Yr gimbal, but these see very limited uses outside of some niche applications, basically anywhere where 3DOF manipulation is needed. Then we get into 4DOF joysticks, which are typically some kind of X-Y manipulator on top of another X-Y manipulator, such as a slider or an analog pointing stick on a typical joystick, though at this point the joystick itself becomes a full-sized hand joystick due to obvious dexterity issues, where the new upper manipulator becomes the thumb's joystick. 5DOF, though rare if they even do exist, would be a combination of the two previous, a 4DOF on top of a Z rotation. Then we get into 6DOF, which are entirely different in that they read both linear and rotated movements of the manipulator in a more complex sensor setup, again due to dexterity limitations; I guess you could take the 5DOF and put a secondary Z rotation under your finger as a bi-directional form of trigger. All of these of which can be considered '3D' because of how we define what 3D is, and that's typically not in clean-cut dimensional axes but rather how we manipulate objects, of which objects are typically 3D, so it's better to define a manipulator by the degrees of freedom it has in total and further refine that by how it achieves those degrees of freedom; for example, the modern gamepad is 5DOF, but it uses two 2-axis joysticks and two triggers on a shared axis. Triggers themselves are typically only a half axis, they can be a full-axis but they're almost never logically implemented as such, probably due to the fact they only move in one direction rather than two; which is why I previously mentioned a bi-directional 'trigger,' which wouldn't be difficult to produce, it's just that nobody really makes them. There's also meaning in the term of 'joystick' modernly as an analog stick, joysticks used to be digital and there's very little implementation of such outside of arcade-like uses, but even then a joystick has to have a return to center mechanism, otherwise it's a continuous stepper which is why dial-like devices should be excluded when talking about a degree of freedom, as a DOF should be free to be manipulated rather than detented to specific steps; of which people that don't know better will probably get confused and just suggest that glorified buttons in the form of a stepped dial should be a 'solution' to adding a degree of freedom, sure our scroll wheel on our mice might work for zooming but we're also actuating that in steps rather in a continual input. At least this is true for manipulated inputs, the term 'degree of freedom' is also dependant on use; and some might argue that digital inputs can be a degree of freedom, but I don't really consider pressing a key of stepped input equal to the amount of what an analog sensor outputs is really a 'degree of freedom,' and I know someone will be pedantic enough to argue that 'it's all digital anyways, because that's how USB communicates, and analog devices use an ADC, etc.,' but that's really beyond this argument considering a stepped input is still one-bit binary whereas an ADC will typically be 8-bit if not 12 or 16, so even by that argument anlaog, whicle digital in the end, is still much more refined. Mice (trackballs included) and trackpads are in a limbo between the two, as they're essentially a vector device, but they're still not 'continuous' in the sense of what an analog sensor reads as, not to mention their physical, and often digital, boundaries.
Input devices are more complicated than just calling something a '3D joystick,' it's better to use more refined terms, and to define how the input device actually reads its input.
.....Its called a waldo. They've been used for decades in animatronics.....
@@TankR No, it's not. Waldos are also for digital puppeteering, not animatronics. Not to mention how large these devices are when, if you had any amount of reading comprehension, the discussed topic is about mouse-sized devices. But given your entirely incorrect use, and incorrect formatting, ellipses, I doubt you're literate enough to understand literally anything within this comment thread.
Here's the main issue, Waldo is just a name for a range of devices, and a Waldo is purpose-built for a specific rig in digital puppetry, whether that puppet itself is digital or physical; again, none of this is on the animatronics side of a physical puppet. A great example of this is the difference between Mr. Volt's (@MrVolt) Wheatley puppet's Waldo and production Waldos such as what Henson studios is doing for their Waldo rigs; the former is purpose built for the various degrees of freedom a specific puppet needs, the latter is more open to multiple puppets but is widely used as only a head rig while alternative puppeteering rigs control the body of the puppets- and here's the thing, the vast majority of Waldos are discretely for head and neck manipulators. Whereas what is being discussed here, stemming from what the original device in the video was as I discussed literally only changing actuator to sensor, i.e. to make a driven output device into a driven input device, is a 6dof manipulator that's small enough in size to be used by a hand on a desktop; I've even mentioned the SpaceMouse and its DIYed alternatives, this should've been your second clue, right after watching the video and presumably having an understanding of the original device. And here's the next thing, how degrees of freedom are used in naming, what the SpaceMouse is, its alternatives, and what I suggested in my original comment are three linear axes and three rotational axes on the same manipulated part, whereas Waldos on the other hand are rarely considered '6dof' in this way, nor are executed as such, very similar to how robot degrees of freedom are otherwise named and calculated; you're comparing apples to oranges, hell at this point it might as well be une comparaison de pommes aux pommes de terre, if you understand French and have enough logical understanding of why it's a bad comparison via naming convention, of which I'm doubtful of you having any capabilities of logical thought, comprehension competency, or even basic understanding period.
The next issue is size, Waldos are absolutely massive and complex. You might as well recreate a Novint Falcon for a 6dof manipulator at this point if you're arguing a Waldo would be of any practical use here. Again, your first clue should've been the fact that the device in the video is driven by voice coils, and you presumably understand how utterly small voice coils can get, given that a lot of IEMs (earbuds) are dynamic drivers, i.e. driven by voice coils. Again, your second clue should've been the relevant devices that have been named dropped throughout this comment thread, even if you don't know the device itself it is otherwise easily searchable; and stemming from this, even if you didn't have relevant knowledge of the named devices, you have analogous knowledge considering they're compared to a computer mouse. And no, Waldos cannot be reasonably sized down, considering they encapsulate your hand and wrist at minimum, then on top of that you also have a stable frame to keep them in place or you have a perch that sits under your forearm. Even if you could size down a Waldo to even properly fit on a desk and integrate into a peripheral ecosystem, it would again akin to be what the size of the Novint Falcon was, if not even large; and again here, I'm thinking you lack any amount of logical thought, comprehension competency, and basic understanding of things.
So, what does this mean? Waldos are nonapplicable for the argument and are irrelevant to the conversation, and they cannot be made to work in a way that makes them applicable to the conversation. You lack any amount of reading comprehension competency, literacy, logical thought, the capability to research anything, and lack even the basic understanding of anything relevant to the conversation nor the basic understanding of what you're even talking about. Perhaps you should educate yourself, and until the point of being educated I would suggest that you sit down and shut up. Your reply is on-par with some of the most misguided, misinformed, and unintelligent horseshit I've read in relation to intelligent conversation. Even out of respect for intelligent conversation, and by extension respect for others' time and intelligence, you could have at least done your due diligence to understand the conversation itself, rather than spewing nonsense that is proven to be irrelevant, especially when you do not inherently understand what you yourself are talking about. You can't even have respect for basic conversation, any intelligible portions of the conversation aside, as your incorrect use of ellipses heavily implies a tone that I, or by extension the other party as well, have no idea of what we're talking about and implies a tone that you are talking down to us as if we're lesser, pauses in speech as how you executed always imply this tone, in verbal speech, in writing, and in literacy uses such as dialog in story writing; and this is a point you can't weasel your way out of. People like you act like they're so intelligent, yet are some of the most unintelligent, act as if you're so highly informed, yet are rather the least informed, then you go off on this twatty behavior such as the tonal implications of how you formatted your reply, yet you're the one that's entirely deserving of being talked down to; you should be embarrassed of yourself, I'm sure others are, so again you should sit down, shut up, educate yourself, learn some very basic life skills such as logical thought and comprehension competency, and learn some basic respect for literally anything you interact with. Given that your account has existed since 2013, that implies that you're easily in your 20s if not older, things like this should not have to be explained to you, I get it if the education system and your parents failed you as a child, but you've failed yourself in your adult life, you were likely a class clown but now you're outright just a clown that provides nothing but dogshit on a silver platter. And on an academic channel, where most people are active in intelligent conversation, no less. And on a comment that's actually speaking of the mechanism itself and having advancements elsewhere, no less. Do better, a cat pissing in sand is more sophisticated than you'll ever be.
This is an excellent production, both the mechanical design and research, as well as the video which was quite comprehensive.
Seriously awesome work. Bridging mathematical, computational, and real world testing is a lot of hard work. Mathematical analysis was awesome, FEA was excellent, and real world validation was great to see. Lots of varying feedback, in the comments, but for what this is, I think your team shot a bullseye!
I love this kind of format, a quick overview of a journal paper. Especially useful for mechatronic systems as it is much easier to understand movement of systems in animations rather than static images. As for the content, In particular I like that you kind of explored/mapped out the entire design space (7:15). That must have been quite computationally intensive, did it take a long time?
0:55 Having seen the T-flex stage in real life is very cool. Not just the folded leaf spring spherical joints are very cool, there's lots of little tricks applied in the actuation as well.
Also, I noticed that the stage/platform of the hexblade is constructed out of a polymer, is this done on purpose to have some compliance in the platform? Or perhaps to make it quick and easy to exchange for other types of end stages?
How much would it cost to use this technology to build a small robot/vibrator for the female market? I'm pretty sure we could make some money with something like this.
We could also train a neural network using electrodes on a few girls to find the best algorithms.
Such an underated channel
here comes anoth underrated insanely great channel
This is fucking awesome. I don't care about any applications of the tech even. Your analysis and validation of the mechanical behavior is impressive enough.
Whoo hooooo voice coils rock! :-)
If you need to stabilize that which cannot be seen with the naked eye, always a great option - love the flexure action of soft robotics.
The Future is NOW.
Thanks for the video!
Excellent presentation - exceptionally clear and comprehensive. Would love to see the MATLAB and SW code and learn from it.
Wow, a ton of work has gone into this research. Thanks for sharing in such a deliciously consumable way!
Excellent level of knowledge and sophistication. Top Notch.
Everything was created in beautiful and most effective way. Why!
Because the creator is The Affectionate and have the knowledge of all things.
He gave us physics to show us how hard and beautiful his creation is, So we love him for giving us all of it.
This is really cool, great job student Yang
Really interesting and well put together video. Thanks
Nice work.
Gave me new points of view for my own dabble in 3D printed flexures.
Very impressive! I was surprised to see you were using aluminum for the flexures; I would have guessed that fatigue would be a concern on those.
had same thought too...
the deformation is likely in the elastic range so fatigue/cold work occurs less - I would guess that's the red colour in the simulation
Great video! I took your class last year Hopkins and this was a great refresher on what I've learned from then. Good job, I like the music choice and the script was interesting and easy to follow (if you are an ME i suppose) :)
Completely agree on the music, and the script was top-notch; even a software fool like me could follow!
This was so fascinating. I especially appreciate the use of voice coils as actuators. Were there development problems or scope issues that you encountered at any point along the process that created more work that you expected?
Incredible work! Nicely done!
Nice Rockwell Retro Encabulator you got there!
I love the design and the use of the matlab script you used to calculate it. My one complaint is that 7075 is poor in fatigue, and this seems like a design ripe for a fatigue failure after a few thousand hours of operation. For such thin flexures there are many steel alloys that would be superior at a minor weight penalty.
CF composite from 3D printed molds?
A print head mounted on one of these platforms could print a line path in non-flat layers that are optimized for physical properties like resonance and flexure.
It might be useful to be able to directly generate the Gcode to FDM print a (fibre reinforced?) PLA/nylon/PC positive for this platform's frame from. Perhaps simulate the FDM line structure as well as the shape.
So, even more maths. If implemented in an online calculator it would need a very long progress bar and probably a PayPal button.
I first saw this project (and your channel) at ASPE last year. Cool to see you make a video on it
This video makes me proud to be a mechanical engineer
Fantastic work and great video.
Very well done. Every design choice was justified quantitatively and rigorously. The fact that the experimental natural frequency corresponded so well with your predictions is very impressive!
What sort of resolution and bandwidth can you achieve with that camera-based sensing system?
Nice work!
Great work and presentation. But why not SI units!?
Great video, awesome work. Still baffled by the sampling rate of the camera array. I would have used a 6DOF piezo sensor, because 6 kS/s for 120hz would be nice to have a good amount of datapoints to look out for harmonics, and it would have been way cheaper.
I don't know why I watched it because I'm a computing scientist, but I liked it.
Okay, first off I see a massive amount of work went into this presentation.
Secondly I love Solid works I don't think that there is a better design tool available.
As to my point: How much weight is acceptable for the printer head or are you constrained to a mirror?
I'd be interested in seeing the FEA stress analysis and fatigue life of the parts made from 7075 Aluminium. I can visually see the displacement so that would be more valuable information than URES.
Awesome! But how to replenish printing supplies?
I understand it somewhat at an intuitive level, but the math is like ancient Greek to me. Pretty fascinating stuff.
Fantastic design and demonstration. I just wondered why the coil isn't fixed and the external cup moves so as to remove any flexure to the driving wires? That would remove any chance of fatigue that may cause failure, however unlikely given the small range of movement? Thanks for the brilliant demo.
Often the coil side is lower mass, so you put it on the moving side to keep your frequency response up.
Awesome and informative video. Gets the blood pumping to get to know more about such mechanism.
Nice! Although I have to wonder, once physically built, how lengthy was the calibration procedure? I'd imagine that any inexact placement of the coils would result in non-orthogonal motion?
An aluminum flexure at space temperatures? Am I missing something?
what are those piston like extenders used in the apparatus
This took an entire mechanical engineering degree to understand
Love this!
How does this positioner eliminates vibrations? Destructive interference?
AYYY Yo that DARPA research is MAD. Got so many ideas I want to test now. Im in my last year of my ME degree but I plan on doing research during summer involving flexure devices for sub-micron precision with vibration.
Incredible.
how do such mechanisms fare in fatigue? Is replacement in orbit due to Aluminum's lack of an Endurance Limit a consideration?
Hello
which motors are you using in this ??
Would love to see the finished product maybe one day …
Very interesting!
Narrator sounds like he's reciting the Turbo Encabulator script.
That is exactly what I was thinking!
I’m curious about the folded flexures which act not just as simple bending elements, but also experience significant torsional deformation in certain movements. At first glance this would seem to make it hard to predict the position of the stage since the connection to the voice coil has multiple degrees of compliance, and deform in different ways depending on what the other voice coils are doing. I guess I’m wondering if there is “crosstalk” between the axes.
Given their thin rectangular cross section, each flexure is far more stiff in the direction of intended motion, and bends very easily in the case of not. If you watch the animations at the start, I hope that makes sense. You're likely talking when motion is not perfectly defined by 3 actuators at once, but I assumed the bending moment would still mostly be in line with the flexure's intended direction of bending, so it wouldn't be too disruptive (since its cross section is like 20 times as thick in one direction than the other, and it only receives a small portion of the load that's off axis in the first place. Hope that makes sense?
@@jasondean37 I get it, but yes, I am talking about the twisting induced by the movement of the other axes. The prototype appears to be open loop, and I think it would be quite hard to create an algorithm which would tell you what voltage the voice coils should have for a given position output. If the voice coils had encoders it would be nearly as difficult. Maybe if the end effector could be directly measured for position feedback it could work.
Woah. I only understood about 65% of that but it's really cool.
Great stuff! I did subscribe even though the music at the end is too too loud.
its interesting, im not trying to knock your work or anything, but it seems like the amount of movement it has, is so tiny, especially considereing the size, amounts of parts, and complexity it is to make. Would it scale to have much greater movement somehow?
The problem the design was meant to solve is explicitly a very small fast & precise area of movement relative to the mechanism. You are asking to solve a different problem.
Is this available in cad? May I have?
Yes go to my Thingiverse account. It should be in the description. If not, it is in other video descriptions
Thank you!
The journal article is behind a paywall...😒
Sorry I can't give it out for free or I will be hit with a copyright infringement :( If you have access to a friend at a university or company that pays journal fees though you can get it for free from them. Good luck.
yes major bummer. thought that being darpa related would make the research open, but maybe not
Thanks a lot Master.
We really love those little electromagnetic pistons dont we?
Can we get the matlab file?
yes please!
The script sounds like the Rockwell turbo encabulator except it all makes perfect sense
I was thinking the same thing. If you just play the audio you'll be asleep before the halfway point.
Can we make a compliant 3D printer?
No dislike means this is good stuff 🙂
Why are there three linear motors for the z axis? Isn't that redundant?
No, there are six independent actuators for the six DOFs. To understand redundancy in actuation and constraint, you can watch the first and last lectures of my compliant mechanism lecture serries.
@@TheFACTsofMechanicalDesign thank you so much for putting those up. I'm only on lecture 7. So far on the FACT complete library of parallel systems I've only found 5 degrees of freedom so I haven't gotten there yet/my brain is too tiny to even understand why 5 degrees of freedom would be necessary.
But in 3d printers they only use one z stepper which is so interesting because on this project it seems like the z axis is in part the mechanism crawling on the part! So I expect that the printer has to incrementally print in the z between crawl steps which is also so beautiful.
In deltas, it feels like some of the goal is to reduce the number and cost of actuators and modern printers only use 3. Of course they've got a lot of extra mass and don't get to play with linear motors this excluding the need for linear actuators and 3d tracking for to the cost. I guess use of more degrees if freedom must have something to do with decoupling the different movements and it must have more capabilities than a parallel linkage in a typical printer.
Anyways, thank you so much for your response and I can't wait to finish the rest of the lectures. I'm sure it will all make sense after I finish. Thank you again so much for putting them up!
We made this in school with an Apple IIe and some legeo in the 1980’s
ELI5: It uses flexy things to move and rotate in almost any way you'd like. Engineers and scientists like that.
It would be interesting to remove the voice coil and instead use toothed magnets (stepper-style) along the non-grounded (currently voice coil) bump stop and then coils on the grounded frame for feedback sensing and a more constrained initiation/execution of movement.
Meanwhile, down on Earth for personal purposes, a 3 degree of freedom flexure-based system that could be easily 3D-printed on typical low-end printers would be wonderful for reducing parts count and (hopefully) wear issues. Well, one can dream, right?
The content of video is very impressive but I am very sorry to say that the background "music" is very annoying.
cool.
could this set of solutions be applied to items sold at the local home depot
I found it strange that they view 7075. I would prefer a hexagon lattice crystalline metal like titanium.
great design! But why the hell do you need this underlying "music" ???
Interesting - the animated model is spinning too fast to effectively see it's mechanisms. slow it down a lot for better meaning.
Did you just read the academic paper out loud as an explanation?
"large range of motion" (looks like it goes 2 inches, maybe)
Seems like a better solution would just be to have 2 rigid bearing mounts to straddle the beam as it is being built, the mounts themselves could be adjusted in very small increments within the rigidity in order to maintain precision. A laser guide would be optimal along the path....and the whole thing would cost far less and take up less space.
my brother what the hell is darpa building up there?? cant imagine that you know or if you’d even be able to tell me, but thats so scary lol
Watching this video makes me miss taking 162b 😭
Still waiting for 162C 👀
Hanzhen harmonic drive gear , over 30 years experience , robot joint , strain wave reducer,
Can somebody build this into a CD player so my CDs stop skipping?
ELM for all of this? That's nuts.
Very cool concept and video :) Insane stuff.
One critique: while it's probably fine for the intended audience (who else would click on a video with this title?), the language was very technical and jargon-y. I think this video could have been improved with some less mentally demanding writing.
The background noises sure make this video hard to understand.
You had me until you mentioned it was for a DARPA project.
man, you really did say a lot of..words. and such.
This is really cool, though I must say the narration is a little turboencabulatory
Ofcourse! Why launch anything when you can just print it? Damn DARPA you've done it again. We went from salted nukes to printed nukes 😂
Эта концепция поможет починить туалет на МКС? Астронавты реально страдают!
anyone seriously using "eigenvalues" in a sentence (without laughing) is worth listening to...
This video reminds me so much of the Retro Encabulator.
ruclips.net/video/RXJKdh1KZ0w/видео.html
Man, I'm so dumb!
The 70's called and they want their sound effects back. Why do people insist on doing shit like this?
Take a breathe man
audio narration is like someone reading form the patent - not understandable
I'll pretend to have understood... 😂
"flexture ... gobldegook ... parameters ... gobldegook ...." etc. Didn't understand a word of it Which part is the flexture? What part is the parameters? Which part is the gobldegook? Not a very edumakational video
Big words...........lots of big words.
The music though. Whyyyy can't concentrate on what you are saying at all with that annoying loop going on. Sigh. Oh well.