One of the most exciting robot news stories of 2009 was the announcement of a $5 million dollar, 5 year international research initiative involving humanoid robots funded by the National Science Foundation. The project pairs a number of Korean and American robotics labs, namely KAIST (Korean Advanced Institute of Science & Technology) with Pennsylvania’s Drexel University, with the goal of closing the gap between Asian and American know-how in humanoid robotics.
This partnership has led to the creation of Jaemi HUBO (aka HUBO 2 / KHR-4), one of the world’s most advanced humanoid robots. While the mechanical design was done by Prof. Jun Ho Oh, the exoskeleton was engineered by Drexel University’s RJ Gross. If you’ve been reading Plastic Pals for the past few months you’ll know all about the robot, but I thought it was time we got to know some of the people working on it. Dan Lofaro has been making trips to South Korea and has helped to build the robot in America on the cheap ($88,000 as opposed to $400,000). As the robot’s primary caretaker, he is working on developing a new control interface and hopes to use the robot to promote science and education. He gives us the goods after the break…
Plastic Pals: Please introduce yourself to our readers.
Dan Lofaro: My name is Daniel M. Lofaro, I am a PhD student at Drexel University in the Drexel Autonomous Systems Lab (DASL) with a concentration in Robotics. In 2008 I received an MSEE in Electrical Engineering in Control Systems. I am one of the care takers of the full sized humanoid robot Jaemi Hubo. I am also a coach for the FIRST Robotics Team 709 from Agnes Irwin in Bryn Mawr, PA.
Plastic Pals: What drew you to robotics, and specifically humanoid robotics?
Dan Lofaro: I have always been a fan of technology, partially due to Bill Nye the Science Guy. The primary reason why I was drawn to robotics instead of another field is because it encompasses many different fields of study including electronics, computer science, mechatronics, etc.. Being in robotics allows me to study many different fields at once and form a functional item. Making a functional robot, something that can deal with the real world all on its own is also one of the main reasons why I love working with robotics. That and MOST importantly ROBOTS ARE AWESOME.
Plastic Pals: I agree! What is the primary focus of your research with the robot?
Dan Lofaro: Human-Humanoid Robot interaction through voice, vision, and mind control (using EEG).
Plastic Pals: From what I understand, you’re working on a control system that combines a non-invasive brain-machine interface (BMI) with situational awareness (dynamic artificial intelligence) and sensor feedback to facilitate controlling such a complex robot. By using this new interface, you could simply command the robot to move forward and it would negotiate stairs without being explicitly told to do so.
What was it like seeing HUBO 2 become the world’s third full-size humanoid robot capable of running?
Dan Lofaro: I am excited by Hubo 2’s success in running. It makes me realize that you do not need hundreds of millions of dollars (referring to ASIMO) to achieve this level of technical excellence. I am proud to be a part of the Hubo team.
Plastic Pals: Why is it important that America get on board with humanoid robotics development?
Dan Lofaro: In 2005, the NSF hosted the “Robots: An Exhibition of U.S. Automations from the Leading Edge of Research” that was a result of a 2-year comprehensive study assessing the state of robotics worldwide. The report compared R&D programs in the U.S., Asia, and Europe. The overwhelming finding was that America leads in areas like medical and military robotics, but is losing this edge and is far behind in others, particularly in robot mobility and humanoids. Indeed Asia leads the world in the electromechanical design of humanoids.
Arguably however, America’s expertise is in machine learning that provides robots with cognition and perception. Without such abilities, humanoids are ill-equipped to understand and perceive the world around them and work in unstructured environments, manipulate objects, and socially interact with people. Current methods for human control of humanoids drastically reduces their overall functionality. This limitation becomes the critical gap that needs to be overcome in order for the next vertical shift in humanoid technology to occur.
Plastic Pals: Have you seen Boston Dynamics’ PETMAN humanoid robot prototype?
Dan Lofaro: I did attend a talk by Dr. Marc Raibert (President of Boston Dynamics since 1992) where he showed their vision for the future of Boston Dynamics. This presentation included a brief talk and video of PETMAN. The extremely impressive parts about PETMAN are:
- the use of heel-to-toe walking (how humans walk)
- the demo is of a BigDog chaise that only has two legs (using basically the same controller)
Being a Controls Engineer I am very fond of the limit cycle form of control that they use for walking on both BigDog and PETMAN – it is closer to how humans walk, as opposed to the normal ZMP walking that Hubo and ASIMO uses.
Dan Lofaro: After I finish my PhD I would like to continue my efforts in bringing technology to younger people. My plan right now is to procure a full sized humanoid and go on tour with it. I would like to do comedy shows, musical concerts, and game shows with the humanoid robot. This way it will bring Hi-Tech to the public in a passive sort of way, as if the robot was just another human. This will allow for greater interaction between humanoid robots and society than there ever has been in the past. After pursuing those plans I hope to find a home at a university that shares my love for robotics and human robot interaction.
Plastic Pals: Looking to the future, what applications do you see for humanoid robots in 10, 20 years?
Dan Lofaro: Humanoid robots are a form of robot that has great potential to make a positive impact on society. Today we have robots that vacuum the floor, cut the lawn or make us a cup of coffee at a single touch of a button. Though these robots are useful the fact remains that they are only able to accomplish one task. You can not ask a Roomba to make you a sandwich or even go up the stairs to get something from the bathroom, in addition no coffee maker that I know of can wash the dishes.
Because humanoid robots are built like humans they can interact in the real world (which is made for humans). A humanoid can walk up stairs or down the street, any where a human can go. If you need a sandwich made, a humanoid can grab bread and meat out of the refrigerator and make you a sandwich, if you need your floor vacuumed then the humanoid will grab the vacuum (the same one that you always use) and vacuum the floor.
Basically anything a human can do a humanoid will be able to do. A humanoid robot will not be the best at doing any task, they are like humans (multi purpose machines). For example construction cranes are not multipurpose machines, they are good at one thing (lifting) which both humans and humanoids can do, just not nearly as well. However when was the last time you saw a construction crane driving a car or cutting the lawn. In my opinion Humanoid Robots will not be feasible for an average consumer to own for more than a decade or two, the price of each unit is much higher then that of a high end car and at the moment the human/world interaction is in it’s infancy.
Plastic Pals: Thanks for taking some time to talk with us. I think I speak for humanoid robot aficionados everywhere when I say we look forward to seeing your future plans in action!
To learn more, head over to the DASL homepage.