Issues encountered during replication of IROS2016 results

[clematisation]

Here are some issues my classmates and I encountered during our replication of your results. We installed the explauto library, and imported the following modules: "environment.py" (https://github.com/flowersteam/Unsupervised_Goal_Space_Learning/blob/master/src/environments.py) and "learning_module.py" (https://github.com/sebastien-forestier/ExplorationAlgorithms/blob/master/learning_module.py). It confused us at first, as we expected these modules to be part of the explauto library. Then, several bugs occurred when we ran the code:

• functions "compute_explo" and "prop_choice" are not defined; they do not seem to be part of explauto either. Where can they be imported from?

• "learning_module.py" calls class "MiscRandomInterest" of module "random.py" (explauto > interest_model), assigning 7 arguments ("conf", "expl_dim", "competence_measure", "win_size", "competence_mode", "k", "progress_mode"), whereas "MiscRandomInterest" takes only 4 arguments ("conf", "expl_dim", "competence_measure", "win_size").