Domain Generalization via Optical Flow: Training a CNN in a Low-Quality Simulation to Detect Obstacles in the Real World
Bibliographic record
Abstract
Many applications in robotics and autonomous systems benefit from machine learning applied to computer vision, but often the acquisition and preparation of data for training is complex and time-consuming. Simulation can significantly reduce the effort and potential risk of data collection, thereby allowing faster prototyping. However, the ability of a data-driven system to generalize from simulated data to the real world is far from obvious and often leading to inconsistent real-world results. This paper demonstrates that some properties of optical flow can be exploited to address this generalization problem. In this work, we train a neural network to detect collisions with simulated optical flow data. Our network, FlowDroNet, is able to correctly predict up to 89 percent of the collisions of a realworld dataset and easily achieves a higher detection accuracy when compared to a network trained on a similar dataset of realworld collisions. We release our code, models and a real-world dataset for collision avoidance as open-source. We also explore the relationship between the complexity of the input information and the ability to generalize to unseen environments, and show that in some situations, optical flow is an interesting tool to bridge the reality gap.
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 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 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".