PRE-SLAM: Persistence Reasoning in Edge-assisted Visual SLAM
Bibliographic record
Abstract
Visual-SLAM systems involve computationally in-tense operations and are challenging to run on embedded devices. One method to alleviate resource constraints is to leverage the edge computing paradigm to offload computationally heavy tasks. Limiting the resource use of Visual-SLAM on a mobile device allows us to deploy such systems on diverse hardware including wearables as well as enable long-term operation. Long-term operation brings other challenges however, such as the need to observe changes in the scene on repeated visits. To address semi-static scenes, there has been some recent work in designing techniques that can filter out these dynamic observations [1] called feature persistence filtering. Recently, such filtering has been demonstrated using Visual-SLAM systems as well [2]. In this work, we introduce PRE-SLAM, which builds upon the edge-assisted Visual-SLAM system, Edge-SLAM [3], to incorporate feature persistence filtering. We revisit the centralized persistence filter architecture and make a series of modifications to allow for dynamic feature filtering in an edge-assisted setting. Using two locally collected datasets, we show how our split persistent filter implementation is comparable with the centralized version in performance, reducing map-point and keyframe retention by 26.6 % and 16.6 % respectively. By filtering out dynamic map-points from the system, we demonstrate an improvement in average localization accuracy by more than 50%. We also demonstrate how incorporating feature persistence filtering into Edge-SLAM retains the key benefits and performance enhance-ments of an edge-assisted Visual-SLAM system, with an added communication overhead of only 500 KB while decreasing overall man size by 8.6 %.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".