IMAC-AgriVLN: Can Agricultural Vision-and-Language Navigation Agents be Aware of Instruction Mistakes?
TL;DR AI
2 min readKey summary
The paper extends agricultural vision-and-language navigation with a new benchmark for faulty instructions.
It finds current agents perform much worse when instructions contain mistakes.
A new IMAC module compares instructions with the front-facing image to detect and correct errors.
The goal is to make agricultural robot navigation more reliable in real-world settings.