Wireless signals are like a light, and if some objects and the surface will be reflected back, the human body is naturally no exception. Different materials will have different influence on wireless signals, and the wireless signals will have special changes after being reflected back from the human body. Researchers in Computer Science and Artificial Intelligence Laboratory (CSAIL) have designed a wireless motion tracking system that captures the reflection of the human body to wireless signals and collect health and behavior data.
This system called Marko broadcasts RF signals from the space, which has a specific change after reflecting the person in the motion, and then analyzes this signal reflection through a specially designed algorithm, and will with an individual Associate, that is, Marko can distinguish from different people based on wireless signals.
In addition, the system can monitor movements within a space that determines a digital specified space. Combining the collected motion information and other related data can be presented that the relationship between personal behavior, people and the environment. You can learn about Marko's working principle in this video.
In order to test and demonstrate, the Massachusetts Institute of Technology has been monitored by patients who have often become anxious because some unknown reasons. This test was carried out in a auxiliary living facility and licensed by the patient's family and nursing staff. After more than a month test, the researchers finally concluded: "For more than a month, Marko has measured the situation in which patients came back and forth in their own independent event space. It is reported that the return is an exciting manifestation. By increasing the number of steps with the visitor log of the patient, we found that the patient has become more excited in several days after the family visits. This test results show that Marko can provide a new Passive tracking means for patient health in the home. "
This system is deployed in a fixed target space, and continuously transmits RF signals, creating a vertical coordinate system and horizontal coordinate system based on reflected signals, and the individual is marked as a bright spot, indicating its position in the three-dimensional space. The horizontal coordinate system defines the general position of the person, the vertical coordinate system determines the feature of height and building.
The system can generate a short-term motion trajectory called "small trajectory" by analyzing the radio packets reflected back in a mobile state (with a frequency of 30 frames per second). In order to separate different individuals and process their motion information, the researchers use a technique called "convolutional neural network", which is a machine learning model commonly used in image recognition applications. The team implements a filter, marked a small circle around each person, and when the body moves, the system can filter to the signal outside the small circle, thereby reducing the amount of data. This signal is combined with signals in the circle, the motion trajectory, and associate a particular radio frequency reflection and a particular individual.
Before the Marko works, the system needs to first collect the information of the target object. The responded person needs to wear a low-power acceleration sensor that is labeled in the measured person's identity information in the reflected RF signal, the first step, the system generates a separate "small trajectory", then pass the algorithm Specific acceleration features match the motion feature. When the user gets up, the measured acceleration information will turn around when the user is turned once, and the acceleration becomes a flat line. The system determines the best matching consistency between acceleration data and small trajectory and labeled these small trajectories to the subject.
In this way, the system can distinguish which reflection signals correspond to which particular measurement object. After collecting the first original data, the subject does not need to wear the sensor. The Marko system is highly efficient and can mark the correct individual identity with accuracy of up to 95%.
The most important potential application scenario of this system may be in a health care mechanism. Marko can passively monitor the interaction between patients, doctors, doctors, and nursing staff to determine whether the patient has accepted drug treatment on time. In addition to health care applications, it can also collect data such as family dynamics, daily work, sleep habits. Marko is a more comfortable and more affordable solution than camera monitoring systems, which provides greater privacy and focusing on motion, sleep, and interactive behavior modes within a particular area and time period.
Despite the advantages of privacy, the team also pointed out that "ensuring technology will not be abused to violate the importance of personal privacy." Because the CSAIL laboratory of MIT is based on the Ministry of National Defense, it is also always received by the Ministry of Defense's fund support. Marko can also be used in military use and anti-terrorism operations. In these cases, ordinary people must be clear that their privacy may be damaged by a certain degree.
This interesting invention has been published in conference papers in the CHI computing system conference.
Our other product:
