analiza aiqbench
This research introduces SUM-AgriVLN, a method that enhances agricultural robot navigation by integrating spatial memory modules to leverage past 3D reconstructions for current decision-making. By treating navigation episodes as connected rather than independent, the approach significantly improves success rates on the A2A benchmark while maintaining high navigational efficiency. The work represents a tangible advancement in applying machine learning and computer vision techniques to autonomous robotics within complex agricultural environments.
SUM-AgriVLN: Spatial Understanding Memory for Agricultural Vision-and-Language Navigation
arXiv:2510.14357v3 Announce Type: replace Abstract: Agricultural robots are emerging as powerful assistants across a wide range of agricultural tasks, nevertheless, they are still heavily relying on manual operations or fixed railways for movement. The A2A benchmark and the AgriVLN method pioneeringly extended Vision-and-Language Navigation (VLN) to the agricultural domain, successfully navigating agricultural robots from starting points to target positions following natural language instructions, while we observed a limitation: In practical agricultural scenarios, users often give repetitive instructions, but AgriVLN treats every instruction as an independent episode, overlooking the potential to use past spatial memories to assist present episodes. To address this limitation, we propose the SUM module, which executes spatial understanding via 3D reconstructions and saves spatial memories via 2D representations from the past, thereby assisting the decision-maker to recall the spatial characteristics of the scenes in the present. We integrate it into the AgriVLN backbone to build the SUM-AgriVLN method. When evaluated on A2A, it effectively improves SR from 0.47 to 0.54 with only slight sacrifice on NE from 2.91 m to 2.93 m, demonstrating the state-of-the-art performance in the agricultural VLN domain. Code: https://github.com/AlexTraveling/SUM-AgriVLN.
Abstract:Agricultural robots are emerging as powerful assistants across a wide range of agricultural tasks, nevertheless, they are still heavily relying on manual operations or fixed railways for movement. The A2A benchmark and the AgriVLN method pioneeringly extended Vision-and-Language Navigation (VLN) to the agricultural domain, successfully navigating agricultural robots from starting points to target positions following natural language instructions, while we observed a limitation: In practical agricultural scenarios, users often give repetitive instructions, but AgriVLN treats every instruction as an independent episode, overlooking the potential to use past spatial memories to assist present episodes. To address this limitation, we propose the SUM module, which executes spatial understanding via 3D reconstructions and saves spatial memories via 2D representations from the past, thereby assisting the decision-maker to recall the spatial characteristics of the scenes in the present. We integrate it into the AgriVLN backbone to build the SUM-AgriVLN method. When evaluated on A2A, it effectively improves SR from 0.47 to 0.54 with only slight sacrifice on NE from 2.91 m to 2.93 m, demonstrating the state-of-the-art performance in the agricultural VLN domain. Code: this https URL.
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2510.14357 [cs.RO]
(or arXiv:2510.14357v3 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2510.14357 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 277 cuvinte
Subiecte conexe
SpaceVLN: A Zero-Shot Vision-and-Language Navigation Agent with Online Spatial Cognitive Memory and Reasoning
arXiv · 9 iun. 2026
SparseNav: Instruction-conditioned Sparse Semantic Perception for Training-Free Vision-Language Navigation
arXiv · 24 sept. 2026
LongSpace: Exploring Long-Horizon Spatial Memory from Perception to Recall in Video
arXiv · 6 iun. 2026
AdaGeoVLN: Selective Geometry Across Representation Depth and Navigation Time for Vision-Language Navigation
arXiv · 17 sept. 2026