Ask Your Question

[RMF] Fault tolerance?

asked 2020-11-13 13:46:26 -0500

achille gravatar image

The Robotics Middleware Framework (RMF) allows for scheduling and traffic management for heterogeneous fleets of robots. Are there any precautions in place for dealing with faults such as a robot going out and suddenly stopping response, (occasional) WiFi deadzones and the like?

If an mobile robot completes a task but ends up in a deadzone, or if the entire network connection goes down for 10 minutes, does that cause an entire operation to go down?

If there are no such precautions in place, what are some approaches to handling that?

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2020-11-17 01:27:52 -0500

Do you have a specific application or scenario in mind, or is this a generic question?

I'll just answer generically: there are several layers to the system. It's expected that network deadzones will periodically result in loss of contact to robots, but RMF (specifically rmf_core) actually is talking to fleet managers, which are often provided by vendors and system integrators, and hence are "already" presumably tuned to deal with dead zones specific to a particular deployment. Presumably they would have already addressed network coverage in areas where tasks are typically completed.

RMF (through its fleet adapters) is paying attention to tasks as they evolve and replanning as necessary to avoid conflicts. In general it tries to be "hands-off" and allow the fleet managers to accomplish their tasks however the fleet manager sees fit, so long as conflicts are not in the foreseeable future. If long outages are expected and normal due to network coverage (or lack thereof) such expectations would need to be built into the fleet manager, which could provide estimated position updates back to rmf_core, even in the absence of robot comms. When the robot pops up again on the network and offers a position update, then the fleet adapter could adjust its predicted path rollout to match reality, and at that point, rmf_core could need to step in and start a negotiation if a conflict is foreseen.

edit flag offensive delete link more


Thanks for your answer. The scenario(s) would be very large factories, warehousing, and industrial logistics centers with hundreds of robots of different types. There's a high cost associated with downtime, calculated in $$ per minute. In those settings it's impossible to guarantee 100% network up-time at all locations, but the robots have to continue the job as best as they can for as long as possible.

achille gravatar image achille  ( 2020-11-17 10:52:20 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools



Asked: 2020-11-13 13:46:26 -0500

Seen: 126 times

Last updated: Nov 17 '20