Robotics Intro

Outline

• This project consists of building a robotic arm with visual perception, and learning to control it through reinforcement and imitation learning
• Here’s the breakdown of the task:

1. LED control
2. Single-servo control
3. Two-link programmatic control
4. Two-link manual control
5. Two-link imitation control
6. Perception
7. Two-link RL control
8. Building the robotic arm
9. Robotic arm programmatic control
10. Robotic arm manual control
11. Robotic arm imitation control
12. Robotic arm RL control
13. Robotic arm sim2real RL control

What is this?

• this is an outline for a project
• the project is to build a robotic arm and control it via reinforcement and imitation learning

Why am I doing this project?

• RL and stochastic dynamic programming have been successful in performing a number of tasks like game playing and robotics
• Most RL (human) learning resources focus on implementing algorithms and applying them in simulation
• Real-world robotics is a key application area of reinforcement learning because it is a case where RL is very likely the best option for learning control policies
• The goal of this project is to demonstrate the application of RL in a robotics task
  • In a manner that people could do by themselves at reasonable cost without external resources like a robotics lab