MaDi: Learning to Mask Distractions for Generalization in Visual Deep Reinforcement Learning
About
The visual world provides an abundance of information, but many input pixels received by agents often contain distracting stimuli. Autonomous agents need the ability to distinguish useful information from task-irrelevant perceptions, enabling them to generalize to unseen environments with new distractions. Existing works approach this problem using data augmentation or large auxiliary networks with additional loss functions. We introduce MaDi, a novel algorithm that learns to mask distractions by the reward signal only. In MaDi, the conventional actor-critic structure of deep reinforcement learning agents is complemented by a small third sibling, the Masker. This lightweight neural network generates a mask to determine what the actor and critic will receive, such that they can focus on learning the task. The masks are created dynamically, depending on the current input. We run experiments on the DeepMind Control Generalization Benchmark, the Distracting Control Suite, and a real UR5 Robotic Arm. Our algorithm improves the agent's focus with useful masks, while its efficient Masker network only adds 0.2% more parameters to the original structure, in contrast to previous work. MaDi consistently achieves generalization results better than or competitive to state-of-the-art methods.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Reinforcement Learning | Procgen (test) | BigFish Return11.96 | 21 | |
| Visual Reinforcement Learning | DMControl Cheetah Run | Episode Return432 | 16 | |
| Visual Reinforcement Learning | DMControl Reacher Easy | Episode Return766 | 16 | |
| Visual Reinforcement Learning | DMControl Walker Walk | Episode Return574 | 16 | |
| Visual Reinforcement Learning | DMControl Ball in cup, Catch | Episode Return884 | 16 | |
| Visual Reinforcement Learning | DMControl Finger, Spin | Episode Return810 | 16 | |
| Visual Reinforcement Learning | DMControl Cartpole, Swingup | Episode Return704 | 16 | |
| Visual Reinforcement Learning | DMControl Walker Run (test) | Environment Reward382 | 5 | |
| Visual Reinforcement Learning | DMControl Pendulum, Swingup (test) | Episode Reward (ER)372 | 5 | |
| Visual Reinforcement Learning | DMControl Hopper, Hop (test) | ER80 | 5 |