[CS391R] Introduction of Robot Learning

Course website

Types of robot automation

• custom-built robots -> human expert programming -> Special-purpose behaviors
• General-purpose robots -> Robot learning -> General-purpose behaviors

What is robot learning?

The study of methods and principles that make robots learn from data

-> Learning is critical for taking robots to the real world.

1. Robot perception: seeing and understanding the physical world
   1. Multimodal understanding
      • 2D object detection
      • Attention architectures
      • 3D data processing
   2. Recursive state estimation
   3. Implicit neural representations
   4. Synthetic data for robot perception
   5. Self-supervised visual learning
   6. Interactive perception
2. Robot decision making: Planning and control of robot behaviors
   1. Reinforcement learning
      • model-free RL
      • model-based RL
      • Inverse RL
   2. Imitation learning
      1. Adversarial IL
      2. Imitation as supervised learning
   3. Task and Motion planning
   4. Hierarchical policy & neural programming