RRT*-enhanced long-horizon path planning for AUV adaptive sampling using a cost valley
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Despite advances in adaptive sampling, existing methods predominantly rely on myopic (greedy) strategies and single-objective criteria, which inadequately balance long-term exploration and exploitation. Moreover, obtaining real-time computations with complex, time-varying models remains challenging. With the goal of effective sampling of oceanographic variables by autonomous underwater vehicles, we propose a long-horizon adaptive sampling system that integrates a flexible cost valley concept with a non-myopic path planner. Our method addresses autonomous navigation within a fixed time frame while adaptively sampling ocean variables and avoiding obstacles, aiming to reduce the expected variability or classification error at river plume fronts. The novelty of our approach lies in combining variance and classification metrics as sampling objectives into a weighted cost surface that guides the vehicle along its minimal-cost path. We implement this concept using a rapidly exploring random trees (RRT*) strategy for non-myopic path planning. Simulation results based on 100 replicates demonstrate differences in traffic flow, root mean square error, variance reduction (VR), and integrated Bernoulli variance (IBV) under various cost weightings for RRT* versus a myopic approach. The equal weight cost valley appears robust, yielding prediction metrics closer to those in extreme IBV or VR-dominant cases. Statistical results further show that RRT*-based planning achieves only slightly better numerical scores than the myopic method—for example, an IBV of 75.76 (SD 7.26) compared to 75.93 (SD 6.4). A 2.5-hour field trial in a Norwegian fjord confirms that the AUV successfully runs the long-horizon adaptive sampling algorithm in real time on its onboard computing units. • Adaptive sampling methods using non-myopic algorithms. • Multiple objectives combined in a cost-valley guiding the sampling. • Field example with autonomous underwater vehicle doing long-horizon path planning in a Norwegian fjord.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it