Home

Research Objective

• Create a bin-picking robot that can pick up objects from a randomly stacked pile.
• This must be done without requiring a prior model of the objects.

Approaches

• Use end-to-end deep reinforcement learning techniques

Intermediate Tasks

• Implement deep neural network on example problems
• Implement deep reinforcement learning on example problems
• Implement end-to-end learning on example problems
• Implement end-to-end learning on robotic grasping problems.