@@MakeYourPet Usually the IK part is precompiled in their products. I have a hiwonder spiderpi and they give you a pretty limited precompiled inverse kinematics lib for python. So you'd have to roll your own stuff to actually fix it. Besides this, Hiwonder stuff is not bad quality wise and if you just need an open platform with which you can get started quickly it's good enough. Besides the proprietary lib it's all open and you can directly write to the drives as well to roll your own. It just runs a native rasperry pi os with some modifications and has a bunch of example projects to run.
At least they have three servos per leg. I agree with you about the inverse kinematics. The main reason I built a hexapod was to figure out the IK math. My first hexapod was really underpowered and its legs segments were much too long. I couldn't tell if my IK was wrong or if there was just too much slop in the hardware. The last hexapod video I posted (years ago) had pretty good IK but it was limited to keeping the body parallel with the ground. I'm working on getting the hexapod to walk with the body in any orientation. This complicates the math a lot. I think I'm going to need to use rotation matrices to keep track of everything. I managed to avoid rotation matrices in my earlier software. I used rotation matrices for my Lego animations. I'm trying to relearn what did back then.
Your hexapod looks very nice. Definitely consider joining the Discord server, as there are many other enthusiasts there building their own hexapods and it is a nice place to share and ask questions.
Hiwonder has made other hexapods before the jethexa. They have created an entire line of robots that run off of raspberry pi, including the Spiderpi, the humanoid Tonypi, and the Boston dynamics-esque quadruped, Puppypi, and more so it's clear that they aren't hexapod oriented like Make your pet or Zenta would be. They advertise in other videos that their programming is open source, and tutorials for the operation of their products show as well. Improving the jet hexa or other hexapods from hi wonder as a project would be interesting.
I agree with everything that you said. Except that just because it is open source, doesn't mean fixing the IK is doable (or easy) for an average buyer. I wonder if they have been building so many robots, why their IK is not correct yet, since that is fundamental for a robotics company. You cannot just put some hardware together and call it a robot. Just my take as a SWE.
@@MakeYourPet You're correct. Most people looking to buy robot kits would be interested in something other than trying to solve a fundamental locomotion problem. To answer your question about how they could have the problem even with all the robots, they designed, out of all the robots Hiwonder has designed, most of the robots that move across terrain do so with Omni directional wheels. There are only three hexapods hiwonder has developed out of fifteen or so. hiwonder is still decently new, so they can create something that improves from past mistakes. I don't consider myself an expert in robotics, so what I have said is purely speculatory. I want to learn more about the process of designing hexapods and other types of robots, and part of the process is constructive criticism. With that being said, I also hope that Hiwonder notices what was pointed out and fixes it.
Hey! I used Hiwonder products previously for some personal projects, and still have some interest in their current products. I think that at this kind of "affordable" price point, the JetHexa is already plenty good especially for educational purposes. Since the jethexa is open source I'm pretty sure if there were such problems it would be relatively easy for the user to solve.
Hi and thank you for your feedback. They may have good products as a company, and I didn't mean to bash the entire brand. I just have to disagree with the "easy to solve" part, just because it is opensource doesn't mean it is easy to fix. Why they haven't fixed it yet if it was so easy? Or probably didn't care about the accuracy that much? In any case it may be a very good high-end toy for the price, but a educational/research grade product as it is advertised? not sure about that.
I don't think that there is bug in the IK algorithm. I think the robot deforms a bit under its weight. Yes, the algorithm does not take this into account, but how could it?
That could be part of it. My robot is made of plastic and deforms a bit as well, and the servos themselves have a good amount of play (dead band). So we really don't expect it to be 100% in this price range. But based on my personal experience, still the margin of error should be much less than what is visible in the video.
I thing it doesn't have to be an dramatic IK error right away! It may also be that the servos in this Video have not been properly calibrated, some variables, lengths or angles were not settet correctly or the rubber is still to slippery. Something like this can often be fixed right at home.
Yes all of those factors (lengths, angles, calibration etc.) should be correct for the IK to work. My point is, they have used an obviously faulty robot in their ad video, which tells me they either didn't notice it, or they just didn't care about the details. Idk which one is worse!
I have noticed that there are no review videos I can find for this unit. I have purchased one to review, and test. Would you be interested in a collaboration?
Indeed that looks a little weird. Its weight is 2.5 kilos, and servos claimed torque is 35 kg/cm, which should be enough not to slip. I've had such issues, but robot was much heavier and servos were much weaker.
I will, if they send me one. It is difficult/not fair to review a product just based on a video. This particular review was based on the obvious issues in their own ad video and made sense.
Based on everything that they claim in the video, it is a very competitive price. But I have no idea about the quality of the parts or the performance of all those features, and what I saw and could tell in that video wasn't very promising. Anything that comes from me is totally free, and you can buy your own parts based on your own preference and budget.
Haha, slipping feet are the first thing that I look for. That demonstration is terrible and really makes the rest of their claims suspect. It's one thing to have the hardware capability but enabling the software that supports it is entirely different. If they can't get the IK working it's doubtful that their fancy lidar navigation, or anything else, is working any better. I'm a hardware guy so I get it but I'm not charging for my designs and I publish my work as a hardware development platform not a commercial product.
:D but the thing is, IK has a definite answer, you solve it once and it will work forever (unlike the other stuff like face detection), and shouldn't be a problem for a company in that scale. It is totally weird that they haven't done that.
haha, either didn't notice, or didn't care about the details. Idk which one is worse honestly! They claim in the video that it is a research grade robot, they should back that up.
I have the jet hexa and tonypi, the jethexa is pretty cool the controller is good as well but the control distance sucks you can only go like 30 40 feet away before it starts losing connection, the opsticl avoidance works on the app but from what I understand the app opsticl avoidance is definitely not the same as the ROS that it has but I still need to install the packages, yes you have to install the system's but they give you instructions to do so, I have not yet done so, but will when I get the time to do so, but the walking it does is fine I never seen mine slip and even if it does it doesn't effect mine it walks, may be I should post a video or something, but I have other robots and a hexapod with a lidar but not a depth camera and that one only cost me 174$ compared to what I payed for the jethexa 1400$ for the advanced kit, so o know these things can be sold for cheap, you just need to know what to use and how to properly use the hardware software, it just takes a lot of time to do for a new be.
![Nuts [YTP] | A Moana parody](http://i.ytimg.com/vi/MnIRAGodAXA/mqdefault.jpg)
I wonder if the code is open source? Perhaps it's possible to find the problem independantly.
I think it is opensource but not sure about it. Yeah that would make a great project actually. Only If they send me one :P
@@MakeYourPet Usually the IK part is precompiled in their products. I have a hiwonder spiderpi and they give you a pretty limited precompiled inverse kinematics lib for python. So you'd have to roll your own stuff to actually fix it. Besides this, Hiwonder stuff is not bad quality wise and if you just need an open platform with which you can get started quickly it's good enough. Besides the proprietary lib it's all open and you can directly write to the drives as well to roll your own. It just runs a native rasperry pi os with some modifications and has a bunch of example projects to run.
I’ve had this hexapod on on my watchlist for a few months now. Figured I would start with a simpler project before spending the money.
At least they have three servos per leg.
I agree with you about the inverse kinematics. The main reason I built a hexapod was to figure out the IK math. My first hexapod was really underpowered and its legs segments were much too long. I couldn't tell if my IK was wrong or if there was just too much slop in the hardware.
The last hexapod video I posted (years ago) had pretty good IK but it was limited to keeping the body parallel with the ground. I'm working on getting the hexapod to walk with the body in any orientation. This complicates the math a lot. I think I'm going to need to use rotation matrices to keep track of everything. I managed to avoid rotation matrices in my earlier software.
I used rotation matrices for my Lego animations. I'm trying to relearn what did back then.
Your hexapod looks very nice. Definitely consider joining the Discord server, as there are many other enthusiasts there building their own hexapods and it is a nice place to share and ask questions.
Hiwonder has made other hexapods before the jethexa. They have created an entire line of robots that run off of raspberry pi, including the Spiderpi, the humanoid Tonypi, and the Boston dynamics-esque quadruped, Puppypi, and more so it's clear that they aren't hexapod oriented like Make your pet or Zenta would be. They advertise in other videos that their programming is open source, and tutorials for the operation of their products show as well. Improving the jet hexa or other hexapods from hi wonder as a project would be interesting.
I agree with everything that you said. Except that just because it is open source, doesn't mean fixing the IK is doable (or easy) for an average buyer. I wonder if they have been building so many robots, why their IK is not correct yet, since that is fundamental for a robotics company. You cannot just put some hardware together and call it a robot. Just my take as a SWE.
@@MakeYourPet You're correct. Most people looking to buy robot kits would be interested in something other than trying to solve a fundamental locomotion problem. To answer your question about how they could have the problem even with all the robots, they designed, out of all the robots Hiwonder has designed, most of the robots that move across terrain do so with Omni directional wheels. There are only three hexapods hiwonder has developed out of fifteen or so. hiwonder is still decently new, so they can create something that improves from past mistakes.
I don't consider myself an expert in robotics, so what I have said is purely speculatory. I want to learn more about the process of designing hexapods and other types of robots, and part of the process is constructive criticism. With that being said, I also hope that Hiwonder notices what was pointed out and fixes it.
Hey! I used Hiwonder products previously for some personal projects, and still have some interest in their current products. I think that at this kind of "affordable" price point, the JetHexa is already plenty good especially for educational purposes. Since the jethexa is open source I'm pretty sure if there were such problems it would be relatively easy for the user to solve.
Hi and thank you for your feedback. They may have good products as a company, and I didn't mean to bash the entire brand.
I just have to disagree with the "easy to solve" part, just because it is opensource doesn't mean it is easy to fix. Why they haven't fixed it yet if it was so easy? Or probably didn't care about the accuracy that much? In any case it may be a very good high-end toy for the price, but a educational/research grade product as it is advertised? not sure about that.
@@MakeYourPet Chinese companies usually don't care so much about these details. If it works good enough for most customers, they will ship it.
Interesting I was wondering how to tell if those cheap hexapods are any good and this is how to know. Thanks for another great video.
I don't think that there is bug in the IK algorithm. I think the robot deforms a bit under its weight. Yes, the algorithm does not take this into account, but how could it?
That could be part of it. My robot is made of plastic and deforms a bit as well, and the servos themselves have a good amount of play (dead band). So we really don't expect it to be 100% in this price range. But based on my personal experience, still the margin of error should be much less than what is visible in the video.
I thing it doesn't have to be an dramatic IK error right away!
It may also be that the servos in this Video have not been properly calibrated, some variables, lengths or angles were not settet correctly or the rubber is still to slippery. Something like this can often be fixed right at home.
Yes all of those factors (lengths, angles, calibration etc.) should be correct for the IK to work. My point is, they have used an obviously faulty robot in their ad video, which tells me they either didn't notice it, or they just didn't care about the details. Idk which one is worse!
I have noticed that there are no review videos I can find for this unit. I have purchased one to review, and test. Would you be interested in a collaboration?
I have been busy lately and don't have much bandwidth for YT these days, but if you do a review definitely drop a link here or in the Discord server.
Indeed that looks a little weird. Its weight is 2.5 kilos, and servos claimed torque is 35 kg/cm, which should be enough not to slip.
I've had such issues, but robot was much heavier and servos were much weaker.
HI great video and helpful to anyone looking to buy this product. Could you please review the HiWonder Jetauto?
I will, if they send me one. It is difficult/not fair to review a product just based on a video. This particular review was based on the obvious issues in their own ad video and made sense.
Could you review the spider pi pro from hiwonder
What about the cost? Does it justify $699? How much does your hexapod cost, just parts only?
Based on everything that they claim in the video, it is a very competitive price. But I have no idea about the quality of the parts or the performance of all those features, and what I saw and could tell in that video wasn't very promising.
Anything that comes from me is totally free, and you can buy your own parts based on your own preference and budget.
Please also review the "ros master x3" robot
If I get one, will you help me with the code? Or maybe do a RUclips tutorial coding a better algorithm.
If they send me one, yeah I will try to fix the code. I will do a IK video soon
Waiting for this, sir.
IK video coming...
The Hexapod is way to expensive and also finding out by this video(thanks a lot!!!) it's not even a finished product.
Can you do the hiwonder robot please
I don't have it unfortunately, but I will be more than happy to do that if they send me one! :D
@@MakeYourPet If they send you one and you'll be able to walk through the code in a video to explain the IK, I'll definitely buy one!
Haha, slipping feet are the first thing that I look for. That demonstration is terrible and really makes the rest of their claims suspect. It's one thing to have the hardware capability but enabling the software that supports it is entirely different. If they can't get the IK working it's doubtful that their fancy lidar navigation, or anything else, is working any better. I'm a hardware guy so I get it but I'm not charging for my designs and I publish my work as a hardware development platform not a commercial product.
Yes, that is exactly what I was thinking. I think I couldn't convey that in the video as well as you said it here.
The fancy lidar navigation is probably just some ros packages thrown together and will work well enough in environments with enough features.
Haha not noticeable for us common folks
:D but the thing is, IK has a definite answer, you solve it once and it will work forever (unlike the other stuff like face detection), and shouldn't be a problem for a company in that scale. It is totally weird that they haven't done that.
@@MakeYourPet maybe they didnt know that's a problem hehe
haha, either didn't notice, or didn't care about the details. Idk which one is worse honestly! They claim in the video that it is a research grade robot, they should back that up.
I have the jet hexa and tonypi, the jethexa is pretty cool the controller is good as well but the control distance sucks you can only go like 30 40 feet away before it starts losing connection, the opsticl avoidance works on the app but from what I understand the app opsticl avoidance is definitely not the same as the ROS that it has but I still need to install the packages, yes you have to install the system's but they give you instructions to do so, I have not yet done so, but will when I get the time to do so, but the walking it does is fine I never seen mine slip and even if it does it doesn't effect mine it walks, may be I should post a video or something, but I have other robots and a hexapod with a lidar but not a depth camera and that one only cost me 174$ compared to what I payed for the jethexa 1400$ for the advanced kit, so o know these things can be sold for cheap, you just need to know what to use and how to properly use the hardware software, it just takes a lot of time to do for a new be.