Humans don't rely strongly on metric information to explore or navigate a new environment, so can we enable robots to do the same? In our newest paper to be presented at this year's #RSS2023, our team from the University of Oxford and New York University explores the idea of metric-free exploration for topological map building. It turns out that with some careful architectural choices and engineering tricks, we can build useful topological maps with great efficiency in terms of sensor usage, training data, and step budget. The topological maps built can then enable strong visual navigation performance. Appreciate the help from all my coauthors and mentors Yuhang He, Yiming Li, Rushi Shah, Chen Feng The paper can be read here: https://lnkd.in/ec3t3S9u We will release all our code next week! #robotics #robot #RSS