A major challenge to assault prevention is that during an assault, victims often do not have an easily accessible way to call for help. Whether calling 911 or using an emergency alert app or device, the user has to press a button in order to call for help. That is often not possible while a violent act is taking place or if a person is unconscious as a result of the assault.
Researchers at UAB have developed a Smart Jewelry Bracelet — currently a prototype — that uses machine learning and a multitude of sensors to analyze a user’s movements in order to detect an assault as it is taking place. Upon detection that the user is in danger, it is programmed to emit a loud beeping sound and to flash red strobe lights in an effort to scare the attacker off and to alert other people who may be nearby. The device then connects to the user’s smartphone via Bluetooth and instantly sends emergency messages and coordinates of the user’s location to emergency personnel and a list of contacts predetermined by the user through a mobile app. The bracelet contains an Adafruit Circuit Playground — a small microcontroller equipped with a gyroscope, accelerometer, temperature and pressure sensors, GPS, and microphones.
The sensors allow the bracelet to collect user activity and vital signs continuously. It can also determine the orientation of the user, for example, whether they are standing or lying down. A machine learning algorithm detects and differentiates the user’s regular movements from unexpected and sudden movements that can be indicative of an assault.
While the bracelet was primarily designed for detecting assault, it can also be used by the elderly or those with disabilities to automatically detect sudden falls or other risky movements. The researchers hope to refine the design and expand the technology to other everyday items worn on the body, such as earrings and shoes.
The bracelet prototype costs less than $40 to create and can be made even cheaper if it is mass produced.