Beyond Memory Limits: A Hierarchical Spatio-Temporal Map Representation for Large-Scale 3D Autonomous Exploration
Beyond Memory Limits: A Hierarchical Spatio-Temporal Map Representation for Large-Scale 3D Autonomous Exploration
This research presents FLAVOR, a hierarchical spatio-temporal framework designed to address memory constraints in autonomous UAV exploration. By confining dense metric geometry to a robot-centric local map and retaining compact global observation and task context, FLAVOR achieves significant reductions in exploration time and map memory footprint compared to baseline methods, both in simulated environments and real-world deployments such as forest and construction site scenarios.
The paper introduces FLAVOR, a hierarchical spatio-temporal map framework for autonomous UAV exploration that reduces memory footprint by retaining dense geometry locally while using compact global observation and task context; reported improvements in exploration time and memory usage are based on simulated scenes and real-world deployments in forest and construction site environments.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
map memory footprint reduction
31.98
View original evidence
ration viewpoint from observations, while the topological route is instantiated and validated against the current local map before execution. Dense-map capacity remains fixed as exploration expands, with exploration-dependent memory growth confined to compact observation and task context.FLAVOR achieves 31.98--41.88% shorter exploration time and a 94--99% smaller map memory footprint than the best baseline that compl
Open source S5map memory footprint reduction
41.88
View original evidence
ration viewpoint from observations, while the topological route is instantiated and validated against the current local map before execution. Dense-map capacity remains fixed as exploration expands, with exploration-dependent memory growth confined to compact observation and task context.FLAVOR achieves 31.98--41.88% shorter exploration time and a 94--99% smaller map memory footprint than the best baseline that compl
Open source S5map memory footprint reduction
94
View original evidence
ration viewpoint from observations, while the topological route is instantiated and validated against the current local map before execution. Dense-map capacity remains fixed as exploration expands, with exploration-dependent memory growth confined to compact observation and task context.FLAVOR achieves 31.98--41.88% shorter exploration time and a 94--99% smaller map memory footprint than the best baseline that compl
Open source S5map memory footprint reduction
99
View original evidence
ration viewpoint from observations, while the topological route is instantiated and validated against the current local map before execution. Dense-map capacity remains fixed as exploration expands, with exploration-dependent memory growth confined to compact observation and task context.FLAVOR achieves 31.98--41.88% shorter exploration time and a 94--99% smaller map memory footprint than the best baseline that compl
Open source S5map memory footprint reduction
92.67
View original evidence
etes all three simulated scenes. In real-world deployments, it reduces this footprint by 92.67% in the forest and 62.42% at the construction site relative to an explicit point-cloud ikd-tree while preserving long-term task continuity under constrained onboard resources.
Open source S6map memory footprint reduction
62.42
View original evidence
etes all three simulated scenes. In real-world deployments, it reduces this footprint by 92.67% in the forest and 62.42% at the construction site relative to an explicit point-cloud ikd-tree while preserving long-term task continuity under constrained onboard resources.
Open source S6
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ration viewpoint from observations, while the topological route is instantiated and validated against the current local map before execution. Dense-map capacity remains fixed as exploration expands, with exploration-dependent memory growth confined to compact observation and task context.FLAVOR achieves 31.98--41.88% shorter exploration time and a 94--99% smaller map memory footprint than the best baseline that compl
Open source S5
etes all three simulated scenes. In real-world deployments, it reduces this footprint by 92.67% in the forest and 62.42% at the construction site relative to an explicit point-cloud ikd-tree while preserving long-term task continuity under constrained onboard resources.
Open source S6
Source：arXiv Robotics — research abstracts · arxiv.org