aiqbench analysis
This research introduces the Strategic Transformer, a lightweight neural architecture designed to solve NP-hard combinatorial optimization problems for real-time navigation in vast indoor environments. By integrating geometric attention biases with local refinement techniques, the model achieves a 94x speedup over traditional meta-heuristics while maintaining bounded-latency inference suitable for onboard robotic deployment. The work advances the state of resource-constrained multi-object navigation by effectively bridging the gap between strategic exploration and tactical exploitation in ultra-large-scale settings.
Strategic Transformer for Resource-Constrained Multi-Object Navigation in Ultra-Large-Scale Environments
arXiv:2609.20227v1 Announce Type: new Abstract: Resource-constrained multi-object navigation in vast indoor environments ($>2,000\text{ m}^2$) poses significant challenges for efficiency and strategic planning. To tackle this, we reformulate the task as a Set Orienteering Problem (SOP), providing an optimization framework under resource constraints where exploitation is governed by the SOP model and exploration is managed by a separate heuristic switcher. Conventional baselines suffer from either rigid planning or myopic behaviors. To overcome these limitations and resolve the NP-hard computational challenges of SOP for real-time navigation, we develop the Strategic Transformer. This lightweight architecture functions as a priority planner that internalizes expert combinatorial logic into a predictable $41.03\text{ ms}$ forward pass while reducing teacher-student information asymmetry. Incorporating geometric attention biases allows the network to effectively model long-range structural dependencies. By coupling the Transformer's macro-plan with a bounded iterative 2-opt local refinement on a capped candidate graph, our framework achieves a $94\times$ speedup compared to heavy meta-heuristics, ensuring bounded-latency inference suitable for onboard deployment. Experiments on ProcTHOR validate that our method successfully bridges the gap between exploration and exploitation, outperforming carefully re-implemented baselines under Progress weighted by Path Length (PPL) and establishing a new benchmark for scalable, resource-constrained navigation.
Abstract:Resource-constrained multi-object navigation in vast indoor environments ($>2,000\text{ m}^2$) poses significant challenges for efficiency and strategic planning. To tackle this, we reformulate the task as a Set Orienteering Problem (SOP), providing an optimization framework under resource constraints where exploitation is governed by the SOP model and exploration is managed by a separate heuristic switcher. Conventional baselines suffer from either rigid planning or myopic behaviors. To overcome these limitations and resolve the NP-hard computational challenges of SOP for real-time navigation, we develop the Strategic Transformer. This lightweight architecture functions as a priority planner that internalizes expert combinatorial logic into a predictable $41.03\text{ ms}$ forward pass while reducing teacher-student information asymmetry. Incorporating geometric attention biases allows the network to effectively model long-range structural dependencies. By coupling the Transformer's macro-plan with a bounded iterative 2-opt local refinement on a capped candidate graph, our framework achieves a $94\times$ speedup compared to heavy meta-heuristics, ensuring bounded-latency inference suitable for onboard deployment. Experiments on ProcTHOR validate that our method successfully bridges the gap between exploration and exploitation, outperforming carefully re-implemented baselines under Progress weighted by Path Length (PPL) and establishing a new benchmark for scalable, resource-constrained navigation.
Robotics (cs.RO)
arXiv:2609.20227 [cs.RO]
(or arXiv:2609.20227v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.20227 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Full text under the publisher's open licence · 284 words
