Oct 13 2017
This summer, a research team headed by Professor Mike Fraser, Asier Marzo and Jess McIntosh from the Bristol Interaction Group (BIG) at the University of Bristol, along with University Hospitals Bristol NHS Foundation Trust (UH Bristol), presented their paper at one of the world’s most renowned conferences on human-computer interfaces, ACM CHI 2017 held in Denver, USA.
Credit: BIG, University of Bristol
Computers are constantly increasing in number and wearable computers, such as smartwatches are becoming more popular. Devices around the home, such as smart thermostats and Wi-Fi light bulbs, are also coming into the market at a good pace. However, existing technology confines the capability to interact with these devices.
Hand gestures have been recommended as an intuitive and easy way of interacting with and manipulating smart devices in various surroundings. For example, a gesture could be used to open or close a window or to dim the lights in the living room. Hand gesture recognition can be realized in many ways, but the placement of a sensor is a big restriction and frequently rules out certain methods. However, with smartwatches becoming the top wearable device this allows sensors to be placed in the watch to sense the movements of the hand.
The research team suggests ultrasonic imaging of the forearm could be used to distinguish hand gestures. Ultrasonic imaging is already applied in medicine, such as pregnancy scans along with tendon and muscle movement, and the Researchers saw the potential for this to be used as a way of comprehending hand movement.
The team applied image processing algorithms and machine learning to categorize muscle movement as gestures. The Researchers also undertook a user study to discover the best sensor placement for this method.
The team’s findings revealed a very high recognition accuracy, and significantly this sensing technique worked well at the wrist, which is perfect as it allows future wearable devices, such as smartwatches, to integrate this ultrasonic method to sense gestures.
With current technologies, there are many practical issues that prevent a small, portable ultrasonic imaging sensor integrated into a smartwatch. Nevertheless, our research is a first step towards what could be the most accurate method for detecting hand gestures in smartwatches.
Jess McIntosh, PhD student, the Department of Computer Science and BIG Group