Real-time reconstruction of fragmented trajectories: An integrated machine learning and behavior-based spatiotemporal framework
Bibliographic record
Abstract
High-quality road user trajectories are essential for various transportation applications. Despite the significant advancement of detection and tracking technologies, observed trajectories often suffer from several issues that impact their applicability, such as intrinsic errors, noise, and fragmentation. This paper introduces a real-time reconstruction framework for road user trajectories, designed to reconstruct coherent trajectories from potentially fragmented segments. The framework begins with processing the raw trajectories to extract several dynamic features such as velocity, acceleration, curvature, and heading. A Random Forest classifier is then utilized to identify trajectory segments likely belonging to the same path. The classifier incorporates the Subsequence Dynamic Time Warping (sDTW) metric and other spatiotemporal features. Next, similar segments are grouped into cohesive clusters where a trajectory reconstruction module merges the identified segments and interpolates missing segments using the Gaussian kernel-based regression. Finally, the reconstructed trajectories are smoothed using integrated wavelet transforms and Savitzky-Golay filters. The framework was trained and validated using trajectory data acquired from the Lyft Level 5 AV dataset. We focused on the reconstruction of pedestrian and cyclist trajectories due to their inherent complexity and unpredictability. Validation results confirmed the accuracy of the different system components as well as the accuracy of the reconstructed trajectories compared to ground truth data (RMSE of 0.1138 m and MAPE of 0.01%). Computational assessments indicate that the framework scales linearly with data size, with optimal performance for real-time applications achieved for 5- to 10-minute windows.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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".