[Question]: visual trajectory in Figure 8 of DualVLN

[leo7611]

Question

Excellent work. I am learning it and encountering some issues as follows:

1. During training, are only vln_ce required from the three datasets: vln_ce, vln_pe, and vln_n1 (vln_n1 is optional and not seen in the training script)? Is vln_pe only used for evaluating isaacsim?
2. How is the visual trajectory in Figure 8 of DualVLN obtained?