Shared autonomy policy fine-tuning and alignment for robotic tasks
Bibliographic record
Abstract
In this paper, we present a comprehensive shared autonomy framework for human-in-the-loop policy fine-tuning and alignment. Our framework integrates policy adapting algorithms on a multi-agent system foundation tailored for human-robot interaction and decision-making arbitration. This strategy is intended for complex, task-oriented robotic tasks that require cognitive-level human-robot interactions. We design short- and long-horizon fine-tuning algorithms to adapt a policy to different operating conditions and human agents. This is accomplished using Bayesian analysis and custom deep reinforcement learning techniques, through various interaction channels strategically placed at different operational points of the system. To showcase the effectiveness of our algorithms, as well as the strength of our framework, we conduct a human user study involving operation of a laboratory robot in a sequence of high-level pick-and-place tasks. The experiments of the study are designed to demonstrate the interplay between different design elements of our framework, such as, interaction channels and multi-horizon fine-tuning algorithms. By laying out careful hypotheses, we employ objective and subjective metrics to measure the effects of shared autonomy design elements on both the system performance and human user satisfaction. Our human user study reveals significant results related to the complex interplay between shared autonomy design elements, the behavior of the algorithms, and core decision-making and arbitration formulation.
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 | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| 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 | 0.002 |
| 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".