Posts by Reckless

want something ultra portable and able to pick up 10lbs

195K views · 3.2K reactions | Reflex Robotics, USA, wheeled mobile chassis robot #humanoid #robot #industrial #intelligent #artificial #knowledge #science #popular | By Amazement Show | Facebook

Reflex Robotics, USA, wheeled mobile chassis robot #humanoid #robot #industrial #intelligent #artificial #knowledge #science #popular

www.facebook.com

I was getting ready to order a humanoid robot for office assembly and home use but think this idea might be better. Would love to hear thoughts

I would imagine the people on linkedin are referring to scara style robots or 6DOF which generally run under $20k. UR runs twice as much. I like cobots but I feel they need to be smaller, lighter and cheaper to be useful in electronics manufacturing. SCARA robots make more sense as they are more accurate and reliable. I am looking forward to cobots replacing coffee, food industry applications if they reduce cost. I feel AI is the magic missing from cobots to complete various tasks/multipurpose robots.

Quote from saberdud

I read the article you posted, and I agree with all of their reasoning for their size of robot. The stepper motors they use pack a lot of punch for their size, but aren't really a good use for industrial sized robots. for larger robots, you want absolutely no backlash, as that puts the weight of the robot on the motors and it introduces a lot more stress. The stepper motors in their robots will not really care about that because the weight of the robot is so low.

I believe their newer models have absolute encoders and are heavier with larger payloads. I am very fascinated with this robot due to its lightweight and trying to see how to reduce weight further. I bought this one with rail kit for packaging operation:

I looked into some other Yamaha robots, will take another look at them.

Quote from MOM

The teach pendant should be a ADTECH TV5300.

Have you checked the different menues whether you can find save/load files?

Yes still no luck.

TV5600-English-User-mannul-20151030170428.pdf

TV5600-dispensing-machine-20151028150441-2.pdf

Quote from SkyeFire

Those Cartesian machines are the "easy mode" of kinematics, though. Closed rather than open, no or poor positional feedback, very low strength, very small work envelope.

Yes, but what I see is a need for cheap robotic 3 axis and 5 axis to cover majority of situations. Dispensing robots run $10-60k and they are variations of this china one which I suspect is only $500 in quantity.

For 5 axis I think Dorna is on a good track but I wish it was even smaller/easier to deploy, similar to mecademic but $2000 instead of $15k. elephant robotics sells similar one for $500 but not reliable.

Just like china is taking over the world by making cheap evs same is going to happen with robots.

I have been tempted to make my own custom super thin, ultra lightweight 4 axis robot similar to this with harmonic drives even at higher cost but previous article posted has scared me off seeing issues they mention (some don't make entire sense to me). Also not sure how long of a reach I can make a 3 axis robot (4th is just on end effector). I would like 500-750mm if I am custom designing. It would resemble this but longer.

Quote from SkyeFire

Everyone's been wanting that for decades. So far, no one's been able to make it work -- that price point simply can't support building a robot with industrial levels of speed, payload, robustness, reliability, accuracy/repeatability, and reach. The circles on that Venn diagram don't all intersect. UR tried to enter the market at the "bargain basement" price point, and took a lot of stick for how their first few robot generations failed early under industrial use rates, even at low payloads. And even UR hasn't been able to get below the $30k mark, IIRC.

If anyone had cracked the problem, I'd expect to see it from at least one of the many new startups stampeding the low-price, low-payload space, but not so far.

I think Epson is close at $7k for their scara. This US startup has done it:

Dobot had one for awhile but they seem to be pivoting.

China is able to produce these type of 3 axis cnc style for $800:

The secret sauce seems to be using stepper motors over harmonic drive:https://dorna.ai/blog/design-pr…a-robotic-arms/

I should have posted to see if anyone was in town and wanted to get some coffee or tour the show together. Show is over, I think it was phenomenal. Every manufacturer was there. Some things I picked up that may be common knowledge for others:

1.) Chinese robots are not trusted because of how much data they are trying to steal. This type of thinking is new to me as I have been in china many times and didn't see it on cyber espionage. I always felt they are way behind in software development. China robot manufacturers are insisting to disconnect their robots from the lan network.

2.) AI robots are the future and everyone will be replaced. Humans are just like the horses that were replaced by cars at the turn of last century.

3.) Humanoid robots are too inefficient. I had a nice meeting with the founder of brooks and this was one topic I brought up. He convinced me they are not the future. He brought up an old failed honda robot and how it needs 2kwh to function and honda gave up. 100 watts/h is better. I don't see Tesla installing them at homes in near future.

4.) Surprising to me that the largest market for robots in 2024 is in the electronics field but I really didn't find any robots that catered to my needs. I like small, light and fast and I prefer cheap cobots (means industrial robots with light curtains/cameras). Electronic manufacturing is on microlevels. Epson has something in the works but I feel their robots are still too big for what I want. I stopped by Mecademia's booth and liked their robots but too expensive for my taste, double epsons prices.

5.) Everyone has 6 dof cobot on the market right now and they all look alike. I have a hard time imagining such large arms being used in day to day environments (coffee shops, mcdonalds, ice cream parlors, etc). No one is distinguishing their robot arms from others. I think there is plenty of room to make lighter cobots.

6.) I learned about harmonic drive and liked their simple robot design but too expensive. I think I am the only one purporting cheap industrial robots, under $5k. I feel like epson's t3 should be under $4k by now after selling a few hundred thousand units. I am having to design my robot from scratch as I am not finding ultra lightweight robots (under 10 lbs). I haven't been able to check if the harmonic motors are cheaper in china.

7.) Everyone was very respectful and helpful that I ran into at the show. Alot of good will. Somehow show is 4 days but feels very short. Too many people to meet. I had atleast 20 manufacturers to visit.

8.) I did learn the reason controllers are not built into robot has to do with power/speed. Higher voltage has to have thicker gauge wire which requires external box.

9.) I stopped by a booth regarding bin picking/3d vision and saw how a cheap elephant robotics mycobot (which I liked for its ultra compact nature) failed within 3 days of the show. No longer on my watch list. If the robot had worked it was a nice tabletop companion.

Next year is in detroit, I am thinking to go, mostly to learn/absorb Robots/automation. Show felt similar to years past, many companies were debuting new robots. I liked the robot dogs/pets running around at the show. Needs more of it to liven up the atmosphere. We need some r2d2's and other robot helpers.

Tried that last night nothing happened. I believe port was previously used to power a small led light, not sure for what.

Quote from MOM

Maybe this is a strange idea.

There is an usb connector at the rear of the robot and two usb connectors (A and B) on the teach pendant.

The usb connector at the rear should work and the usb connector on the teach panel only works if an usb cable for printers (A and B) is connected between rear and teach pendant

so basically find firewire type cable hook it to robot and then put usb drive into b on teach pendant? I will try that but it seems super bizarre to me. need to find one of those old printer cables.

Quote from SkyeFire

...eh? MS and *nix both support FAT32. And FAT16 should still be supported by both, although less so as file sizes get bigger.

Anyway, replacing the teach pendant strikes me as unnecessarily overcomplicated for a first attempt -- it would require completely reverse-engineering the pendant/mainboard communications protocol from scratch, just to start.

If that machine has a regular serial port (sometimes just a set of unoccupied solder pads on the mainboard PCB), that would suggest the board is made to accept commands over that port, and finding a reference to that interface should be much easier.

If you really want to replace the pendant, you're going to need to start by tapping into those control lines and testing what all 15 of them do, for which button presses. I'm going to hazard a guess that at least four of those wires are dedicated as a USB pass-through from the main board, but you'll need to identify them. Two of those breakout connectors from your second post would be a good place to start -- wire them up as a pass-through, then tap in with an oscilloscope or logic analyzer. And settle down to collect a lot of data. Those lines could be a mix of serial and discrete signals, some dedicated to the controls, others to the LCD display.

I agree, need to avoid reverse engineering. My #1 preference is using hand teach to load files via usb, second preference is hooking it to laptop with cameras for verification.

Quote from MOM

The teach pendant should be a ADTECH TV5300.

Have you checked the different menues whether you can find save/load files?

Yes, I have tried my best to load a usb file. I tried both fat16 windows and unix. It doesn't find usb drive or show it.

So I was changing my encoder battery and now not getting DC power. Is there anyone who remotely fixes robots? This is a brooks preciseplace 3 axis robot. Its supposed to have mean time between failures of 50k hours.

WhatsApp Video 2024-02-29 at 6.31.33 PM.mp4

You and me both. But hand teaching is more common then you think but generally its done like an array, put 50 locations in and program offsets. Part of the issue is china uses unix file systems as they are free, whereas most of us use microsoft file formats, fat32 vs fat16.

I did find a cool company with modern touch screen hand teach interfaces for $60 but they have a different communication connector: https://en.coolmay.com/ProductDetail.…=36&Terminal=41

Quote from MOM

Picture, schematics and movie do not match.

What manufacturer and model number do you actually have?

I have the one in first picture called jarvis which I am unable to locate manufacturer. The youtube video was a later model but essentially same. The unix based hand teach does not recognize a usb when I insert it.

Quote from panic mode

how are you loading data right now?

Currently trying to switch to pc before hand teaching 3000 locations per panel (100+ pcbs). It has array function so would need to manually program 100 locations but feel loading a csv file will make my life easier in long run. Also want to add a camera for verification.

Quote from MOM

Your 15 pin connector has no pin for pc communication.

On the lower left of front panel is a com connector.

You would need to check with documentation or home page how this connector may be used

Yes, hand teach connects to com connector. This is how it operates:

I have found this on the internet but not sure how to interface to pc