How to solve "Invalid input argument type or size such as observation, reward, isdone or loggedSignals." error? (Reinforcement Learning Toolbox)
7 views (last 30 days)
Show older comments
Lymperis Perakis
on 11 Jul 2019
Answered: Drew Davis
on 23 Jul 2019
I did an implementation of the 2D Bin Packing Problem using the Reinforcement Toolbox, where i defined my environment and agent. When I try to train my agent I get the error "Invalid input argument type or size such as observation, reward, isdone or loggedSignals.". I checked the output variables of my step function and it seems, that they have valid values. When i manually give the Actions to my environment, it gives the expected output and proceeds without problem. So I guess there is a problem with the agent (Neural Network)? Attached you can find a .zip file containing all the functions needed. The script is the environment.m.
0 Comments
Accepted Answer
Drew Davis
on 23 Jul 2019
In your reset function and step function change:
LoggedSignal.State = {[Blocks];[Bintable]};
to
LoggedSignal.State = {[Blocks],[Bintable]};
The vertical cell array isn't concatenating nicely with another cell array as part of training.
Hope this helps
Drew
0 Comments
More Answers (0)
See Also
Categories
Find more on Environments in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!