For the Fall 2013 Intro to Robotics course (EE125) at UC Berkeley, Aaron Bestick (one of the GSIs for the course) was looking for a way to get his students engaged in some real world problem solving. The course was designed for folks from a wide array of engineering disciplines, from computer scientists and mechanical engineers to bioengineers, so he needed a platform that would be straight forward enough to learn quickly, but flexible and powerful enough to be used for relevant investigation. It turns out the Baxter Research Robot was a perfect fit! The marching orders were simple, the students were asked to combine perception, planning, and action to solve a real world demand in a non-trivial way using the concepts they were learning in the classroom. The results were four fantastic projects using Baxter to demonstrate a working knowledge of the coursework.
Baxter plugs in
This team used image recognition, planning, and motion to train Baxter to plug an extension cord into a wall outlet. Nice work on using a simple method to differentiate between the top and the bottom socket!
Controlling Baxter with a Leap motion controller
Imagine Baxter as a marionette!
Keeping a neat work space
Baxter may not be Rosie from the Jetson’s but this team was able to train the robot to recognize objects that have been moved out of place, and move them back to where they belong.
Baxter, please go sort the recycling
It’s a dirty job, but someone has to do it! This group was able to train Baxter to pick up trash and drop it into a can.
About the Author
Brian is new to the world of robotics, but since watching Short Circuit in ‘86 has been obsessed with finding a robot best friend. Baxter is that robot. Working with the brilliant researchers on the Baxter Research Robot has taught him that we are only just scratching the surface on showing off what a compliant and collaborative robot can do.