Coming up with the algorithm, computing power and robotics needed to make a household robot interact with its surrounding and the people in the house effectively is sometime away, but each breakthrough gets researchers that much closer to having C-3PO’s running our homes with a British accent. Integrating a robot into a household is an extremely complicated process. It involves a powerful computer with the ability to process changes in its environment every second of everyday and make the right decision.
The act of picking an object and moving it from one place to another is simple enough if it is the same object. Robots in industrial applications pick an object and move it thousands of times a day without fail, but this is because it was programmed to pick a specific object. If you place a circular object where there was supposed to be a square object, the robot stops working.
A house has many different objects with many geometrical shapes and sizes as well as different locations. This presents a dilemma for a robot if it doesn’t have the right algorithms programmed in it to make these decisions and the dexterity to perform these functions.
The students at the Learning and Intelligent System Group at MIT’s Computer Science and Artificial Intelligence Laboratory will present two papers that show how robots can use lateral thinking to make up for their physical limitations at the 2013 IEEE International Conference on Robotics and Automation in Germany.
The papers will be addressing the act of picking and placing objects, which sounds like a simple enough task until you start breaking down what is involved. Achieving even the simplest task in undefined areas requires an integrated approach to merge together perception, estimation, geometric reasoning, symbolic task planning and control.
The students are trying to find the best solution for every possible problem the robot will be facing every day. If you take a 10 second scenario from what takes place in the kitchen during breakfast for a family of four, objects are picked up and placed many times by each member of the family. A household robot designed to clean up after they have left the house is going to be confronted with picking and placing many items without spilling and breaking any of them.
The algorithms in the papers created by Jennifer Barry, a PhD student and Annie Holladay, an MIT senior major in electrical engineering and computer science, is a work in progress which allows for modifications through application programming interfaces for addressing different parameters concerning new objects and the physical behavior of said objects. The goal is for the robot to learn its environment and make decisions based on the knowledge it has acquired.
We will eventually get there, but it is many baby steps taken by brilliant researchers around the world that will make it possible.
Edited by Brooke Neuman