Unitree S288 as an alternative actuator

[Eric-Lee]

Hello,

I’m interested in building a Microban but the XL330-M288-T are hard to get right now, at least in the US. Robotis tells me they are backordered until November. I'm exploring the Unitree S288 as an alternative. The S288 seems like a good candidate since it looks very similar to the XL330-M288-T. The stall torque and no-load speed characteristics look similar enough and gear ratio, dimensions, connector placement, and weight are about the same too. The differences I have identified are:

• Mechanical: Existing 3D parts probably need slight modifications to support differences in body, fasteners, horns, etc.
• Power: The S288 lists 6.4-12.6V operating range (12.6 V recommended), so I think you would want to swap the battery, holder, charger, and BMS with 3S components.
• Communication Interface: The S288 uses half-duplex TTL with a 6Mbps baud rate. They support 15 device IDs per bus so you would need two buses. I think the Pi HAT only supports 1 bus? Alternatively, Unitree sells a single-bus serial to USB adapter, so you might be able to add a small USB splitter and plug both buses directly into the Pi.
• Data protocol: Not sure about this part. S288 uses Unitree's own communication protocol (link to spec) with potentially different control and feedback parameters.

So I am guessing there would also need to be major software modifications to support the new actuator and communication protocol and for simulation, a new actuator model would need to be created and policies need to be retrained.

I'm a beginner so I'm hoping to get some feedback from someone with more experience. Does this seem like a realistic substitution? What are the other complications or barriers that I didn't identify? How difficult would it be to create an accurate simulation with the new actuator?

Thank you!

[MarcDcls]

Hi! Sorry for the delay in getting back to you. Thank you for your interest in building a Microban and for the time invested in sourcing alternative motors.

In principle, swapping the actuators while keeping the core mechanical architecture is totally doable. Actually, @AI-FanGe is exploring it with Feetech STS3215 motors in this discussion if you want to take a look!

However, since you mentioned being a beginner, I have to warn you that this will be quite a bit of work. I can confirm that all the blocking points you identified seem valid to me. On top of the mechanical tweaks, the power upgrade, and retraining the RL policies, I see two other issues:

• The fixed 6 Mbps communication: The Unitree documentation mentions a fixed 6 Mbps baud rate. I am really not sure that the Pi HAT and the Pi Zero itself can reliably handle such a high frequency.
• The hardware interface: Rewriting the low-level motor driver to support Unitree's custom serial protocol from scratch can be quite complex. I personally don't have experience with these specific motors, so I can't estimate the difficulty of the task.

Taking on the mechanical tweaks, the electronic changes, and writing a custom hardware interface is a fantastic engineering project, just be aware of the amount of work it implies!

[Eric-Lee]

Thanks for the honest reply. I am up for the challenge, but the S288 cost after shipping + tax ends up more than the original entire project BOM... I saw that Feetech also came out with a low cost XL330-M288-T competitor: HD-1910-C001 which looks like an even closer fit for the original design. I think I will try my luck with this one first