Evaluation of Techniques for Sim2Real Reinforcement Learning
Bibliographic record
Abstract
Reinforcement learning (RL) has demonstrated promising results in transferring learned policies from simulation to real-world environments. However, inconsistencies and discrepancies between the two environments cause a negative transfer. The phenomenon is commonly known as the “reality gap.” The reality gap prevents learned policies from generalizing to the physical environment. This paper aims to evaluate techniques to improve sim2real learning and bridge the reality gap using RL. For this research, a 3-DOF Stewart Platform was built virtually and physically. The goal of the platform was to guide and balance the marble towards the center of the Stewart platform. Custom API was created to induce noise, manipulate in-game physics, dynamics, and lighting conditions, and perform domain randomization to improve generalization. Two RL algorithms; Q-Learning and Actor-Critic were implemented to train the agent and to evaluate the performance in bridging the reality gap. This paper outlines the techniques utilized to create noise, domain randomization, perform training, results, and observations. Overall, the obtained results show the effectiveness of domain randomization and inducing noise during the agents' learning process. Additionally, the findings provide valuable insights into implementing sim2real RL algorithms 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 | 0.006 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
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".