This paper proposes to train an ObjectNav policy that generalises to unseen environments by using a modular system that classifies objects and builds an episodic semantic map, which it is uses to explore the environment based on the object category, building upon the hierarchical method in "Learning to explore using Active Neural SLAM". The method achieved SOTA performance on the 2020 CVPR Object Goal Navigation Habitat Challenge. Interestingly, the policy, trained on Gibson and MP3D, has been transferred and deployed in a real robot, with some success. While the initial reviews were mixed (9, 7, 4, 5), the reviewers converged on (8, 7, 6, 6), agreeing during discussion that the paper deserved to be accepted. Based on the reviews, I recommend this paper for acceptance as a spotlight or poster presentation. Please note that reviewers R2 and R4 have raised several issues that are not fully addressed by the rebuttal and that should be addressed in the final version (in particular, standard deviations and discussions about the final position of the agent in relatively small houses). As a side note, the authors did a commendable job of acknowledging privacy concerns in datasets based on peoples' homes. As reviewer R4 suggests, and while additional ethical concerns about bumping into humans are rarely addressed in the robot community, mentioning this would be a good addition. I would also disagree with R1's request to release the code at submission rather than after acceptance - this requirement typically penalises large projects, and promotes the release of functional but poorly written and badly documented code (as the experiments are barely wrapped up) rather than code that effectively helps the scientific community. Code release is significant work in itself.