Udacity Robotics: Deconstructing the Pick and Place Scenario

Picking and placing an object is something a human takes for granted.

Imagine yourself on a nice Sunday afternoon, finally cooking that delicious recipe you have been waiting to try for a long time. You have your recipe handy, have cut your veggies, prepared your spices, washed your rice, and are ready to start cooking.

Picture credit: http://lumdimsum.com

According to the recipe, first you need to heat some oil in a pan, and sauté some onions. But you can’t actually do that unless you identify and locate those beautiful finely chopped onions among 20 other ingredients.

Using your advanced human vision technique, you scan the options, and finally identify and locate the onions. What next? Simple—you pick up the bowl containing the onions and transfer them into the pot. In reality though, it’s not that simple at all. Your brain actually needs to: 1) calculate an optimal collision-free path from the original location of the onions to the cooking pot, 2) Inverse Kinematically control multiple joints in your arm, and 3) maintain body balance with the newly added bowl to your otherwise stable system—all of this while avoiding knocking over the oil jar (and the other ingredients!). Not so simple, right? Now, try teaching this to a robot!

Kuka KR210 6-DOF arm picking up an object from the shelf

The truth is, a robot regularly encounters similar situations where it has to locate a specific object in a cluttered environment and then move it to a different location without disturbing other objects. So it’s important to understand how this happens.

Over the course of the Udacity Robotics Nanodegree program, we will cover all aspects of solving this problem—from perception, to planning, to execution.

All so that you can enjoy your Sundays while your robot cooks that delicious meal for you!