Summary of Deep Reinforcement Learning with Symmetric Data Augmentation Applied For Aircraft Lateral Attitude Tracking Control, by Yifei Li and Erik-jan Van Kampen
Deep reinforcement learning with symmetric data augmentation applied for aircraft lateral attitude tracking control
by Yifei Li, Erik-jan van Kampen
First submitted to arxiv on: 13 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper proposes two Reinforcement Learning (RL) algorithms that leverage environmental symmetry to improve state transition prediction and control policy optimization. The first algorithm, Deep Deterministic Policy Gradient with Symmetric Data Augmentation (DDPG-SDA), uses symmetric data augmentation to enrich the dataset of a Markov Decision Process (MDP). The second algorithm, DDPG-SCA, incorporates an extra critic network trained independently using the augmented dataset. A two-step approximate policy iteration method is used to integrate training for the two critic networks and one actor network. Simulation results demonstrate enhanced sample efficiency and tracking performance of the developed algorithms in an aircraft lateral tracking control task. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper develops two new Reinforcement Learning (RL) algorithms that work better with symmetrical systems. The first algorithm, DDPG-SDA, makes the training data more diverse by using symmetry to generate new samples. The second algorithm, DDPG-SCA, has an extra critic network that also uses this symmetry to learn from the data. By combining these two networks and using a special training method, the algorithms can predict states and control policies more efficiently. |
Keywords
» Artificial intelligence » Data augmentation » Optimization » Reinforcement learning » Tracking
