Morag Tohamy
The Problem:
We start with a rectangular 2D surface. In the center, there is a robotic arm with 2 joints. Each joint allows movement of 360 degrees in both directions. Additionaly, a ball spawns in a random location on the surface.
The Goal:
We have to make the tip of the robot arm stabilize on the ball.
The Solution
First, in order to build a control system (What is a control system?), we had to acheive knowledge of the system’s dynamics. Since that model was not known to us, we used an ANN to learn it. The ANN learned the model and emulated it for us which allowed us to analyze the dynamics, and apply control system algorithms on the robotic system (That included RHC and ILQR ) to achieve the desired result – get the tip of the robotic arm to stabilize on the ball.
Source
