Event
A New Method for Keyframe Management in Dense Mapping for Memory-Constrained Robots
1 reports · 1 sources
Huichang Yun, Jaeyeul Kim, and Seungho Yoo propose a keyframe management method for dense mapping aimed at memory-constrained robots; it directly limits the memory occupied by the resident map and decides keyframe admission and release according to semantic spatial units such as building rooms and corridors. The authors report that in every run on two NVIDIA Jetson boards the memory budget was never exceeded, with model throughput loss of at most 0.8%; without this method, the operating system terminates the model or the map. The authors report that under this budget, in the evaluated environments, admission by semantic region localizes 6–9 percentage points more query frames than geometric partitioning by the number of matched regions, roughly twice that of RTAB-Map retrieval, and also yields lower ATE. This work has been submitted to ICRA 2027.
Report timeline
- Keyframe Management for Dense Maps on Memory-Constrained Robots
The authors propose a keyframe management method for dense maps on memory-constrained robots that directly limits the memory occupied by the resident map and admits and releases keyframes by semantic spatial units such as building rooms and corridors. Across every run on two NVIDIA Jetson boards, the memory budget was never exceeded and model throughput loss was at most 0.8%, whereas without the method the operating system terminates the model or the map. Under that budget, admission by semantic zone localized 6-9 percentage points more query frames than geometric partitioning by the number of matched regions, roughly twice that of RTAB-Map retrieval, with lower ATE.
arXiv cs.RO ·