Shared Autonomy for Proximal Teaching
DOI: 10.1109/hri61500.2025.10973807
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of personalized motor skill instruction, specifically focusing on how autonomous systems can act as effective teachers rather than just assistants. The authors identify a gap in existing shared autonomy research, which often fails to model how assistance interacts with individual student abilities to determine optimal teaching strategies. Inspired by the educational psychology concept of the Zone of Proximal Development (ZPD)—the gap between what a learner can do independently and with assistance—the paper proposes Z-COACH, a framework that uses shared autonomy to identify which sub-skills are most "learnable" for a specific student at a given time. The goal is to provide targeted coaching that improves interpretable task sub-skills, such as braking or steering, rather than generic performance metrics. The methodology involves a formal framework for leveraging shared autonomy for both student modeling and coaching. Z-COACH utilizes an unsupervised skill discovery algorithm called CompILE, modified with weak supervision from noisy language annotations to ensure the discovered skills are human-interpretable. The system estimates a student’s ZPD by comparing their performance with and without shared autonomy assistance across these identified skills. The experimental design features a user study with 50 novice participants learning high-performance racing in the CARLA simulator, using the Thunderhill Raceway Park track. Participants underwent baseline trials, assisted trials using either strong or weak shared autonomy for modeling, and a five-minute practice session where they received either no assistance or skill-targeted shared autonomy (SKILL SA) based on the ZPD estimation. The results demonstrate that Z-COACH effectively identifies which skills each student should prioritize for practice. Participants who received coaching from Z-COACH showed significant improvements in driving time, behavior, and smoothness compared to those who practiced independently for the same duration. The study found that using strong shared autonomy for student modeling produced results more aligned with expert coaching strategies. By overlaying trajectories, the authors showed that students receiving Z-COACH generally learned smoother racing lines than the self-practice baseline. The system successfully mapped noisy expert feedback to interpretable skill clusters (e.g., braking, throttle, steering), allowing for precise, skill-focused interventions. The significance of this work lies in its demonstration that semi-autonomous capabilities can serve as adaptive teaching tools. By explicitly modeling the Zone of Proximal Development through shared autonomy, the framework moves beyond simple assistance to provide personalized, curriculum-based instruction. This approach addresses the risk of skill degradation due to over-reliance on automation by ensuring assistance is tailored to the learner’s current developmental level. The findings suggest that shared autonomy can be leveraged to design effective learning curricula for complex motor control tasks, offering a scalable solution for specialized training where human experts are scarce.
Key finding
Students receiving targeted coaching via the Z-COACH shared autonomy framework showed significant improvements in driving time, behavior, and smoothness compared to a self-practice baseline.
Methodology
simulator
Sample size: 50
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via author_sweep_intake on 2026-05-28.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-28 |
| archive | success | canonical_url | — | — | 1 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | — | — | — | 1 | 2026-05-28 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Theoretical Contribution: computational model