Even though humans have been walking, running, exercising, and playing sports since pre-historic times, we do not fully understand the subtleties of motion that separate the bodies that get injured and wear out from those that don't. Moreover, we haven't determined what separates the elite athletes from the rest of us. In the last hundred and fifty years, scientists have used principles from mechanical engineering to try to find the answers to these questions and others, creating the field known as biomechanics. While this approach cannot completely explain injury and performance, it may be able to address aspects of these questions that other approaches cannot answer.
By viewing the body as a mechanical system, we can attempt to find mechanical causes for injuries, such as experiencing higher forces in the knee. Perhaps my grandmother experiences abnormally high forces in her knee, which caused it to wear out and require a total knee replacement. Perhaps your grandmother manages to keep the forces in her knee unusually low, so she has been able to run thousands of miles without injury. We can also search for mechanical reasons for elite sports performance, such as using less energy to achieve the same motion.
As technology advances, our ability to more closely analyze the forces experienced and energy expended by the human body is constantly increasing. The end goal of my research is to provide a better tool for performing biomechanical analysis of human motion, namely a video-based motion tracking system that allows much greater flexibility than current systems to analyze human activity in an arbitrary environment without restrictions.
The current state of the art biomechanical analysis of human motion consists of very specialized equipment operated within a laboratory. In a typical running experiment, a subject is asked to run across the room multiple times with reflective markers attached to the skin, while his motions are recorded on specialized video equipment for later computer analysis. This equipment, however, has several limitations. The video cameras we use have special lights and electronics attached to them so that they only see the reflective markers, not the actual body. These markers are tedious and time-consuming to apply and can feel awkward to the subject. In addition, subjects must wear markers on the hip, forcing their clothing to be bunched up in order to get it out of the way. Even if a subject doesn't feel uncomfortable running with bunched-up shorts and taped-on markers, the laboratory is not large enough to allow for a runner to get into his/her rhythm. When a subject feels awkward or can't get into rhythm, he/she may not be able to move naturally, which calls into question the validity of the data we collect.
Moreover, to collect data for the proper force calculations, we have embedded a plate in the floor that measures the force as the subject steps on it. Any trial where the subject does not step on the plate properly must be thrown out. Also, the cameras are too expensive to allow any throwing activities in the lab. While this laboratory setup is sufficient for recording and analyzing slow activities, such as walking, eliminating these shortcomings would make for a system which is much more useful for analyzing many more activities.
The system our research group is developing attempts to eliminate the shortcomings mentioned above. Instead of finding and following individual points, the software we are writing automatically identifies and follows whole body parts, such as the upper arm or the thigh, in a series of images from a standard full-color video recording. Using this method, we don't need to use those cumbersome reflective markers; however, the subject does need to wear tight, but comfortable, clothing. The cameras we use are much less specialized, and therefore more portable and less expensive. Once a subject has been recorded performing an activity from multiple camera views, we match a three-dimensional model to these views to track the motion of every body part, including the arms, torso and legs.
Even if the topic of interest is the dynamics of knee motion, we track every body segment to make it possible to calculate the forces experienced at the knee without the use of a force plate. With assumptions about the mass and inertial properties of each body segment, the software uses the laws of dynamics to evaluate the forces and movements in all of the joints at every instant in time. My work focuses on the identification of the parts of the body from the video images and the automated matching of the three-dimensional model to the images. Accomplishing these tasks consists of developing algorithms and implementing them into software. At present, I am able to identify body parts and am currently working on the problem of matching the model to what is seen.
By moving to less specialized cameras and eliminating the need for a force plate, it should be possible to take this system anywhere. For example, the system could be set up at a baseball diamond to study pitching, in a factory to study the causes of repetitive stress injuries, or at a nursing home to test seniors without requiring a visit to a doctor's office. At any of these locations, the subjects can move naturally because they are unencumbered by markers or a special environment.
By allowing biomechanical analysis or natural movement anywhere, it will be possible to study more people doing more different activities and to gain a much better understanding about injuries and performance than ever before. We will be able to help many more people like my grandmother avoid the deterioration of their bodies so that they can be active through much more of their lives, perhaps even run 50 marathons.
|Modified 15 January 2003 * Contact Us|