Aligning Safe Construction Robot Actions with Human Preferences
Bibliographic record
Abstract
The use of robots has gained increasing interest in the construction industry due to their abilities to improve efficiency, reduce waste, and release humans from labor-intensive tasks. However, deploying robots during construction raises serious concerns regarding workplace safety, especially during human-robot collaboration. Simply put, construction robots must learn safe behaviors when collaborating with workers. Unfortunately, safe robot behaviors on construction sites are hard to define due to the complexity and variety of construction tasks and the constantly changing site environment. To tackle this issue, this paper introduces a novel framework that aligns robot behaviors with actual human preferences regarding safe robot actions. The framework first introduces an online preference labeling tool that allows human experts to choose between safe and unsafe human-robot collaboration from image pairs collected in a lab. Next, we trained a deep learning model to rank various robot behaviors according to human preferences. In other words, images that are deemed safer are ranked higher than those that are deemed unsafe by human labelers. Our model was examined for the tool handover task, which is a commonly seen human-robot collaboration task onsite. Experiments showed that the trained model exhibited high precision in identifying safe robot behaviors and could recover the global ranking of different robot behaviors from the perspective of safety, therefore contributing to the enhancement of safe human-robot collaboration on-site.
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) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.001 |
| 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".