Dan Lofaro

Biped Locomotion:

Motivation:
Every day there are more and more people on the Earth. With a world population of almost 7 billion the need for products has increased dramatically. The advent of the robot allowed a single worker to be more efficient at their single task. Though robots make specific tasks easier they are often only designed to do one thing. These tasks include cutting material with a plasma torch like the Kawasaki FS06L Plasma Cutting Robot, patrolling the sky over Iraq like the US Air Force’s MQ-1 Predator, or simply vacuuming a floor like iRobot’s Roomba. Though all of these tasks are important and each robot is only able to do their task. The Roomba could not pick up a plasma torch and cut something and a Predator would not be a good vacuum cleaner.

Figure 1: Kawasaki’s FS06L Robot—Plasma Cutting	Figure 2: US Air Force’s MQ-1 Predator	Figure 3: iRobot’s Roomba

Being limited to doing a single task is not optimal because it means that we will need to make a different robot for every task we want to have completed. For example if you wanted to have your floor cleaned automatically and a smoothie made for you all automatically you would need a separate robot for each. In addition most of the time for a given robot to function the surrounding needs to be modified, such as adding an invisible wall for the Roomba, or bolting the Kawasaki FS06L Plasma Cutting Robot to the ground in order for it to function properly.

One of the answers to this problem of having to have a single robot for each task you want done is having a single robot for every task you want to have completed. This robot will have to be able to navigate every where a human can so the surroundings do not need to be modified and it should be able to use the same tools that humans use. For example if you want the robot to cut the lawn it should just go to the lawnmower that you already own and operate yourself and the robot should use that to cut the lawn. If it is a smoothie you desire then it should be able to go and get fruit out of the refrigerator and operate the blender just like we do as humans.

Due to the fact that the tasks that we wish the robot to do are all human, and the world we want it to navigate is a world made for humans, it makes sense that the robot made to do these tasks emulate us to some degree. From this idea comes the humanoid robot. A humanoid robot simply means that it is a robot that moves like a human. It will walk on two legs and use two hands just like we do.

Currently there are a few robots designed to move like humans. Honda’s Asimo, and KAIST’s Hubo are two of these Humanoid robots. Each robot has approximately 40 degrees of freedom, including individual finger control as well as full pan, tilt, and roll on each of the hips and shoulders.

Figure 4: Honda’s Asimo	Figure 5: KAIST’s Albert Hubo

The first step in making a humanoid robot able to interact in the human world is to have it take it’s first steps, i.e. walk. Currently this is the place where humanoid robotic research is at, learning how to walk. The current method used in both Asimo and Hubo for walking is using a fixed trajectory with zero moment point, ZMP, control. This means that the robot always has its center of mass over the small area in which it is stable. In order to ensure stability the robot needs to keep it’s knees bent when walking in order to always have full control over the center of mass of the robot as well as ensure it does not hit any singularities. Though this method works it is about half as efficient as straight leg walking. It is important to note that we, as humans, walk with our legs straight, i.e. straight leg walking. Do to the fact that you can not guarantee stability using ZMP control and fixed trajectories another methods needs to be found that will allow the robot to walk more efficiently and like a human.

Topic 1: Biped Walking using Limit Cycle Control
The use of a novel form of control called limit cycle control has proven to work very well on the quadruped robot named Big Dog made by Boston Dynamics. In short the idea of this form of control is to manipulate the dynamics of the system through the use of various controls to ensure the robot falls in to a stable limit cycle. This way no matter what disturbance occurs the system will always be attracted back to it stable limit cycle.

Figure 6: Boston Dynamic’s Big Dog

Topic 2: Study of the Effect of Biped Locomotion as it Pertains to Animals with Tails
Though humans have been around for a few hundred thousand years there was a species that came far before us that also walked on two legs, the dinosaurs. The notorious Tyrannosaurus Rex, or T-Rex, and the vicious Veloso Raptor are two examples of some biped dinosaurs which also had significant tails when compared to their size. It was less than a century ago that paleontologists just assumed that the dinosaurs dragged their tails. This was shown to be false in the mid part of the 20th century by paleontologists showing that the vast majority of the preserved track marks made by dinosaurs found did not have any markings indicating that they were dragging their tails. This means that their tails were there for a reason.

Figure 7: Tyrannosaurus Rex, a Biped Dinosaur with a Massive Tail Compared to it’s Body Size

The purpose of this study will be to see how the addition of a controllable tail to a biped system will affect the dynamics of the given biped walking system.