first action before reading first observation ?

[vincentberenz]

robot_interfaces/include/robot_interfaces/robot_backend.hpp

Line 119 in a738a3a

while (!destructor_was_called_ &&

I find it counter intuitive that the system waits for a first action before starting, as possibly the first action may depend on the robot state/observation.

My program was hanging forever and it took me a while to find that this was because I start my loop with:

robot_interfaces::TimeIndex time_index = ri_frontend_.get_current_timeindex();

which hanged forever as no first action had been sent yet.

[luator]

I agree (I actually had the same issue once). I think the best solution would be to already provide observations immediately and only start sending actions once the first one is provided by the user.

[vincentberenz]

@luator should I go ahead and work on this ? (want to make sure nobody else is currently doing this)

[wumanu]

we should probably talk about it first, felix and i had some discussion and it is not really clear what is the best solution, there are some subtleties. Il giorno mer 13 nov 2019 alle ore 10:31 Vincent Berenz < [email protected]> ha scritto:

…

@luator <https://github.com/luator> should I go ahead and work on this ? (want to make sure nobody else is currently doing this) — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#10?email_source=notifications&email_token=AAO6EOUL6C72FBWT25RI63TQTPCQPA5CNFSM4JFNOND2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED5PKXI#issuecomment-553317725>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAO6EOVFAIXRJZKC7CFV6UTQTPCQPANCNFSM4JFNONDQ> .

[luator]

To have this documented here: We decided in an offline-discussion to try the following solution:

• As soon as the backend is started, it starts adding observations to the time series.
• The action time series stay empty in the beginning and only start "running" once the first action is sent by the user. It starts with the proper non-zero timeindex so that it is in sync with the observations in the same way as it is now.

I can take a look into the code to get an idea how much effort it will be to implement this.

[wumanu]

that would be great!