Right now we are only logging sensor values when the tracer is started. I would like to switch to an "log everything" by default setup. That is, record the 10 minutes over the last 10 robot experiment runs. This is in case the robot does some unexpected behavior / unexpected error shows up.
For this, it might be useful to separate out a separate "logger process" besides the "control" and "hardware" process.
Ok, we could have a process monitoring the shared memory.
I will have a tracer_real_time_over_shared_memory implemented in order to allow RAI to plot online anyway.
So I guess this goes in the same direction.
Right now we are only logging sensor values when the tracer is started. I would like to switch to an "log everything" by default setup. That is, record the 10 minutes over the last 10 robot experiment runs. This is in case the robot does some unexpected behavior / unexpected error shows up.
For this, it might be useful to separate out a separate "logger process" besides the "control" and "hardware" process.