Question on force control for CRX robots

[garyraise]

Hi all, we're developing a robotics solution for assembly tasks. We currently using UR arms, but we're looking for cobots with longer reach and higher payload, which is why we're considering using the CRX-25. From reading through the documentation of this driver as well as the experimental packages, I don't believe there is any existing support for specifying force commands to the FANUC controller. Would anyone have any ideas on how challenging this would be to implement, and whether we could actually use an external computer to control FANUC arms to perform peg-in-hole search algorithms?

[gavanderhoorn]

Technically, this is off-topic for this forum/issue tracker: it's a general usage question about FANUC robots. Not something directly connected to how we might be using/controlling them here.

Having written that:

I don't believe there is any existing support for specifying force commands to the FANUC controller [..]

Unless something has changed recently, none of FANUC's controllers have any form of publicly accessible external interfaces to any form of "force control" AFAIK. Or at least, if by force control you mean: control, in real-time (meaning: high bandwidth, low latency) the forces and torques exerted by the EEF/TCP.

So the problem is not so much there are no packages/projects (ROS or otherwise) offering support, it's available interfaces.

The only real-time interface FANUC has would be Stream Motion (J519), which is strictly position-based (and difficult to buy, at least outside the US). Remote Motion Interface (R912) might be considered an external motion interface, but the performance achievable with that is questionable (see #339 fi).

FANUC robots typically allow for you to configure Force Schedules, which are used for things like "move along this surface while maintaining a 10N force in the Z direction". These schedules are turned on and off around motion instructions, but can't be dynamically updated. That's a form of force control, but not what I'd think of when I read/hear the term.

We currently using UR arms [..]

IIRC, URs also don't really offer force-control in the way I described it above. In a way, they also have 'schedules', but it's called force mode I believe. It's typically used to enable a controlled form of compliance in one or more axes.

Would anyone have any ideas on how challenging this would be to implement, and whether we could actually use an external computer to control FANUC arms to perform peg-in-hole search algorithms?

this would probably depend on what you mean exactly by "use an external computer to control".

Conduct the complete search, while controlling a robot purely using EEF forces & torques? I don't believe that'd be possible (in any meaningful way).

Remotely set some parameters and then have a TP program perform the search, possibly using a force schedule (or perhaps soft float)? Doable, but would you still call that "external control"?

---

Note: I'm not an expert here, and I don't have experience with all FANUC's options. Just a regular user. I'd suggest reaching out to your FANUC branch office for a more authoritative answer.

[garyraise]

@gavanderhoorn Thank you for the quick reply, and I do realize this is a bit off-topic. We did reach out to FANUC already but we were not given much detailed information. We've been able to use the UR interface successfully for the most part, but it definitely limits what types of motions we can specify. I think that since it seems very difficult to dynamically change the force exerted by a FANUC arm even in a scheduled setting, this may not be a great solution for us. Thank you so much for the insight.