Learning from imperfect demonstrations in a surgical training task
Bibliographic record
Abstract
Robotic surgery offers several advantages over traditional techniques, including improved precision, greater consistency, and enhanced dexterity. Learning from demonstrations (LfD) is a promising approach for transferring expert skills to robots, thereby alleviating clinicians’ physical workload. However, a major challenge in surgical robotics is that demonstration data often includes suboptimal or failed behaviors due to human error, fatigue, or the inherent complexity of surgical tasks. Discarding such imperfect data results in the loss of valuable information and hinders the scalability of data-driven surgical skill acquisition. In this work, we propose a novel LfD optimization framework capable of learning from a broad spectrum of demonstrations—including successful, suboptimal, and failed attempts. Our method employs a dual probabilistic modeling strategy to encode demonstrations and formulates a multi-objective optimization problem under novel problem conditions to find an optimal reproduction. We validate our approach on the standard ring-and-rail task, a representative surgical training task requiring high-precision and dexterous manipulation. Real-world experiments using the da Vinci Research Kit (dVRK) show that, even in the presence of failure cases within the demonstration set, our method produces optimized trajectories that enable the patient-side manipulator to successfully guide the ring along the curved wire without contact. These results demonstrate the robustness and effectiveness of our approach in learning from imperfect data, underscoring its potential for real-world deployment in robot-assisted surgery.
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 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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 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".