Can we rebirth AX 12 with X Series specs

dynamixel
robotics

#1

All,
Robotics has created lot of very good servos and the most fevorite I suppose will be AX 12 after MX 64 and X Series. Ax is a really good servo and half a dozzon of robots were build on top of it. Is this possible to upgrade AX to X series specs keeping same 3 d shape? I mean introduction of 30kg cm torque and 12 bit contactless absolute encoder and aluminium body?
NicoX


#2

Thanks guys for liking my post. If Robotics takes this seriously … as we already have XL series than the champion robots like Bioloid GP or Bioloid Premium can be re build using state of art Servos. New AX 12/18 can be brought along with X Series with same outer shape to fit the existing frames and features like current based torque control etc. Please let us know.


(Ko Youngjun) #3

Thank you for your attention to our products. We are happy to be able to receive some opinions about DYNAMIXEL. I’m one DYNAMIXEL developer of our company. We are considering your opinions. However, it is not a problem that can only be determined by our team. we need to gather the opinions of business, production and QC teams. I expect if AX12/18 is released with high spec, the price will be raised.

P.S. our company name is not Robotics but ROBOTIS. it was come from “Robot is …”


(Robert) #4

Hi Ko,
Just a quick question about the AX-12A.
Are you planning to release a firmware update so that AX-12A can also use the Version 2 protocol?
Regards,
Robert


#5

Thanks for considering the fact.
I build my bot around AX 12 and this far I have pushed without PID. Please have a look at this with Odroid XU4 as SOC and USB2AX as servo controller.


Now I need your help as I love my GP.

I feel business can find a better impact as with Maxon motors with torque of 30-35 kgcm ,PID and current based torque control etc feature in AX 12 form factor has awesome impact.

  1. All already available Bioloid frames brackets can be reused and old awesome robots can be rebirth without any effort. Old users does not need to upgrade to new frames as of OP3 etc and they can get all top end research pieces in Bioloid parts also.
  2. Who does not want to create a DARWIN LC (actually DARWIN OP3) who already has Bioloid GP like me.
  3. Personally I feel GP is the best performing bot having AX as mussels.
    When few years back DARWIN LC was seen in some place.
    https://m.youtube.com/watch?v=amiUhtSrHLI

Last of I am sorry for the typo… it will be Robotis … auto correction duhh :frowning:


#6

Please let us know what Robotis thinks about the proposal.:grinning:


(Ko Youngjun) #7

Hi Robert,
I’m wondering what you want for Version 2 protocol. Are you want that the control table is same but just the packet is V2? or Do you want something other function in protocol V2?


(Ko Youngjun) #8

Hi NicoX
We don’t have any project about your proposal yet. But, we still listening the other user opinions about your proposal.


(Robert) #9

Hi Ko,
Thank you for your reply.
The main reason for wanting the V2 protocol for AX-12A is that I am also using both AX-12A and XM430 servos in the same robot arm (with the XM430s near the End Effector).
To use both types of servos to their full capacity I would need to use two contollers - one for the AX-12A in V1 and a second one for the XM430 in V2.
If the firmware for AX-12A servos was also in protocol V2 then everything can be controlled from the one controller.

Best Regards,
Robert


(Leon Ryuwoon Jung) #10

@Becrux

Hi,
so you are going to use AX and XM at the same time, but what you are facing on is the protocol version incompatibility?
I’m wondering if you are using official DynamixelSDK as a high level control system, and so the USB2AX is used for TTL level serial communication.

First of all, official DynamixelSDK gives “Protocol Combined” test source code. This means, you can control on both of protocols in same daisy-chained system. Although, I think that not USB2AX but USB2Dynamixel2 is compatible with official DynamixelSDK.
At second, XM430 supports both of protocols. So, what you have to do before run those is just change the protocol version of XM430. AX and XM430 then will be banded in protocol 1.0.
At third is that I just thought that perhaps you are going to use protocol2.0-specific functions like syncread or bulkwrite? In this case, unfortunately, there are nothing that can be guaranteed to be done. @kyj might be telling you about this story.


#11

Not to repeat, but if Robotis upgrade AX series to X Series electronics with much higher torque etc, users like me and him can utilise, mix and match and create much better performing robots.
Just an openion !!!


#12

Hi @NicoX

I totally agree with you that it would be great if they had an update for these servos to support at least a subset of the Protocol 2 support. I am not sure how much of it that will fit into the processors of the AX, but would be fun to find out!

For what it is worth, I have been playing around with supporting both AX and the XL320 servos on the same buss. So far I have done this both using Robotis DynamixelSDK on the OpenCM9.04 as well as some updates to my own library still WIP (https://github.com/KurtE/BioloidSerial/tree/DXL-Protocol-2-Test)

What I have been doing some playing with is to make a version of the Trossen Robotics PhantomX hexapod, using XL-320 servos instead of the AX-12 or AX-18. So far I have only 2 legs built with the XL320 servos, but I have done some testing with the hexapod with the 4 other legs still being AX servos… A version of the Phoenix code setup to run with combination of servos using OpenCM… is up at: https://github.com/KurtE/Open_CM904/tree/WIP_Support_XL320_Servos

At some point I will also migrate some of this code into my Linux code base, which I have played with on (Odroid, RPI, UP, …). When I do so I will probably try it with the USB2AX (I have a few of them), but will probably most likely start off with Teensy 3.x or OpenCM board …


(Leon Ryuwoon Jung) #13

@NicoX @Kurt

I totally agree that LTS is the good thing.
Though, due to lack of memory, it seems like can’t handle the protocol 2.0 as well as the functions of protocol 2.0 (sync read, etc.).
Actually we can provide solution for using protocol 1.0 motor and protocol 2.0 motor at the same time via DynamixelSDK, but…


#14

Thanks for the response!! I really like it’s not only me, even users like @Kurt, @kyj @Leon_Ryuwoon_Jung are also interested in this topic.
I personally feel there is a wish list going on for some users who want AX series rebirth :slight_smile:

Please let me know if I can bring in any help to community.


#15

Hi @NicoX - As I mentioned, I agree that having the AX12 servos support Protocol 2 would be nice, but as @Leon_Ryuwoon_Jung mentioned, I doubt that the smaller/slower processor in these servos will be able to support the full Protocol 2.

Also assuming you wish the Servos would work like the XL servos, that you can switch between Protocol 1 and Protocol 2, my guess is this could be even harder to get things to fit.

I can imagine there would be issues both for ROM and RAM. I don’t remember which AVR processor is, I think maybe something like an Atmega8A? Which would only have 8K of program space and maybe 1K of RAM.

So there are probably several steps one could maybe take::

a) Maybe support the new Protocol Message: That is the FF FF FD 0 … format, I would think probably doable unless completely out of memory…

b) Updated Registers: The AX-12 has maybe 49 registers. the XL430-W250 can go up to over 600, which I don’t think the AX could handle. So would have to decide which subset? Probably remove all(most) of indirect registers?

c) My guess is that the AX servos don’t have the horsepower to replace the simple Goal position/Speed with the different modes of XL servos…

d) New types of commands. Already supports syncWrite, not sure if again has space or power for Sync Read or other newer commands.

So then the question is, how much would we gain if they came out with a pretty limited version of Protocol 2?

For me, there are two approaches that currently work:

a) Tell all of your new XL servos to communicate using protocol 1. This works. I have not tried it, but have seen others that have.

b) Setup your code to handle both Protocol 1 and Protocol 2. This is what I have been playing with. I have version of my Phantom Phoenix Hexapod code that will at startup time try to locate all of the configured servos, it will check for them using Protocol 1… Then for all of the ones it does not find will try with Protocol 2… It remembers which protocol was used and the program then deals with it. Like when it tries to output the new positions for all of the servos, it splits up the update into those which are Protocol 1 and generates the servo positions (angle to value) and outputs them, then does likewise for Protocol 2… And in the hopeful case that the robot was built with all protocol 1 or 2 servos it can bypass one or the other… (Likewise on OpenCM 9.04 boards the initialization code checks both Serial1 and Serial3…)…

As for me, I hope that over time, all of my stuff will be transitioned over to these newer servos.

Hopefully Robotis will help more with making it easier to do this. Example: the US store almost never has any of the back case idlers in stock for these XL430-W250 servos.

The prices of the brackets made specifically for these servos appear to be pretty limited and expensive. BUT, for example if your are wanting to use some of the brackets for AX servos, example metal C Bracket from Trossen robotics (http://www.trossenrobotics.com/bioloid-metal-f2-bracket), Assuming you can actually get the idler, it is easy to use these brackets. You simply need to use a simple spacer between the bracket and the servo horns… I made my own that were 1.5mm thick and use one on both sides, to make up the difference in servo widths…

Sorry if I am a little off topic.


#16

Thanks @Kurt for writing back. I did not wish to subset a great protocol like protocol 2 and somehow pour into a slower processor like AX12. Instead I will if we can think otherwise.

If you investigate the CAD design for servo horn and idler for both X Series and AX you will find they are similar in dimension. Robotis can create a new X series clone whose outer body, screw holes, horn idler and their thickness matches 100% with AX. Job done right.

Don’t change internals, keep the electronics motor and protocol same as protocol 2, just give a mechanical facelift to be a clone of AX12.

Sorry if I was not sure about my point.


#17

@NicoX :

You mean you would like something like the http://support.robotis.com/en/product/actuator/dynamixel/mx_series/mx-12w.htm servo that is geared and powered more like the AX-12 servo It has a better processor in it. Don’t know if they have a protocol 2 version for it or not… Or if it is going to be continued…

As for me. I personally will go more with the XL series servos as, they finally got away from those … nuts that are a pain in the … Actually would prefer they come out with an XL430… That was somewhere between the XL430-250 and the next ones up, for maybe in the range of $100

Now it would have been great if maybe the cases wee setup to be easier to adapt them to the parts with existing AX servo hole patterns, but hopefully as more and more people adapt to these newer servos, it won’t be as much of an issue in a year or so… Keeping fingers crossed