Running the project and getting results.

[NevilleKitala]

This is just to get deeper instruction for the execution of these project files.

[NevilleKitala]

I just needed to confirm some information on running this project that was outside the previously created issues.

1. Do I need to have sphero.py from the sphero_ros package to see the effects of sphero_res_learner.. files on my sphero robot.
2. all the reservoir information has been created in smp_base and just referred to in the sphero_res_learner.. files.

[x75]

1. no, we are sending geometry_msgs.msg.Twist on /cmd_vel and expecting sensor_msgs.msg.Imu
   and nav_msgs.msg.Odometry on /imu and /odom. the learner requires an action, cmd_vel here and a response channel, the sensors for learning something. the lag needs to be set to match the system you are using

[x75]

2. smp_base is the place for all basic models and algorithms

[NevilleKitala]

Ah Ok, that makes sense. I am reading through your report and the code at the moment. Why did you choose to use negative quadratic distance as your performance measure for? Why not just the quadratic distance?

[NevilleKitala]

I am currently trying to generate the reports being used by res_learner_1D_analyse.py . I can't seem to find log files that are being generated for the analyse code to use. Please advice.

[x75]

hey, sorry it took me a bit to get back

performance measures: feel free to plug in perf of your choice. neg-square is motivated by keeping the model compatible to differential hebbian learning and binary reward on improvement of perf. if we want proximity can take inverse distance aka * -1.

let me check for the reports

[x75]

logfiles: it's a bit dumb, logfile is only saved to file once at the end of an episode in the line

s.savelogs(timestamp, filename)

when i run it i get

...
publishing zn
z, zn [[0.]] [[0.21694477]]
target_str jumping_sign
logs saved to sphero_res_learner_1D/log-learner-20190522-171256-N-500-eta-0.001000-theta-0.200000-g-0.999000-target-jumping_sign
target_str jumping_sign
ending

the files are in sphero_res_learner_1D/ or whereever args.datadir points to.

if you terminate the script before it reaches len_episode no logs will be written