← all datasets

4,565-hour crowd-sourced dataset

Emerging
1papers using it
2026first seen

The '4,565-hour crowd-sourced dataset' is a collection of navigation data gathered from 161 locations across 35 countries, used to evaluate the generalization of imitation-learned navigation policies for point goal navigation in unknown environments.

🔎 Find this dataset

Papers using 4,565-hour crowd-sourced dataset (1)

4,565-hour crowd-sourced dataset — datasets — robotics