 For the last two decades, we've been focusing on information technology to deliver the information faster and more efficiently and more effectively. But now we're entering a new era where physical services can play an essential role in our society. A good example for that is healthcare systems for the health care. So let's take a look at technology. In most factories, car factories, robots are doing much of work. They're already better than humans. They're faster. They're much more precise. They're much more consistent without being tired. But they achieve these amazing performance by being very, very rigid and just following this pre-programmed pattern over and over again without having much intelligence or sense of touch. So when it comes to complex tasks, only humans can do. Robots are not yet capable of doing this complex manipulation task. So if you use this manufacturing robot technology for other purposes, this is what's going to happen. This is footage from DARPA Robotics Challenge. Even though they look like a superhero, it's nothing like a manufacturing robot. Their component technology is exactly the same. They're very, very rigid. I call that moving sculpture. And then they only follow this optimized trajectory, the motion planning, and it doesn't have enough adaptations or dynamic flexibility. This is how animals do in contrast. Animals, they're not as rigid. They're not even nearly as accurate as a machine. But they're much more accurate than other things. Balance, energy absorption, flexibility. Each step, there's so much complex interaction with the ground we have to handle. So in our lab at MIT, we've been focusing on developing different type of machine that can handle this high impact and physical interaction. This is MIT CHIRA tool running on the treadmill. We had to design an entire system from scratch. I'm going to show you the component. It can run as fast as me at this point. Not using full power, though. The limitation is not the power. The limitation is actually the algorithm and the sensor. We haven't put a vision sensor here yet. So because of the adaptability, it can run rough terrain, same algorithm running on the treadmill, and then outside. It can turn. Not as great. We're going to show you a better one a little bit later. Now, it has a sensor in front. It's a laser sensor. It can detect the obstacle and jump over obstacle autonomously. But the most important part is actually landing. That's where this harsh dynamic physical interaction is happening, and then manufacturing technology cannot handle this. It can jump over about 40 centimeter obstacle at this point. It can jump actually much higher, but it doesn't know how to land. As I said, jumping is much easier. Landing is hard. So the key idea here is not only powerful and forceful, but also you have to be very flexible. That's how we are. We're very strong, but we're very, very flexible. So this is much more close to our arms and muscle than those manufacturing technology. And if you try to do this with the manufacturing robot, it won't go anywhere. It's like a sculpture. And that's why these manufacturing robots are only following fixed pattern. It cannot be adaptable. This doesn't come free. We had to design our own electric motors, and we had to design our own transmission and develop our own power electronics to achieve this very special feature. So you might confuse that robots are walking around, but the same technology is required to walk around at the same time, manipulation, and interact with the physical world, which could be patient. I would like to introduce our latest version. This is a Cheetah 3. It looks different from now because there's so much difference between biology and engineering world. For example, we have a 600 muscle. The dog has 600 muscle. We only can afford to have 12 motors. So we had to design very differently, and range of motion is very critical. This is without camera, without having no knowledge about the stairs. It's all about sense of touch because our robots are designed to be able to interact. So it's all about where the foot hit the ground, and then robot knows where the foot is hitting, and then it'll adapt accordingly. We have a camera right now where it starts integrating, but the vision is giving you only a very rough idea. It doesn't give you all the detail. We are not relying on accuracy of vision. So adaptability, again, this is not about accuracy. It's about intelligence, about balance, intelligence about physical interaction. So this is only possible when you have a right components and right paradigm. So I want you to think about this case, how challenging each contact to help a patient, an old person, it's a weak person, not even a paralyzed person, to move from a bed to a wheelchair. This is daily life. It's so challenging once you get to lose that muscle strength. But each contact is so delicate and complex. It's even more than just like a regular running around. We have a lot to do. We have a lot to work on this area. So I want to finish with a quote from Rodney Brooks. Intelligence is determined by the dynamics of interaction with the world. This is an area very underestimated and ignored, and then we have so much to do. Thank you very much.