Activity Tracking
July 11, 2013, Motion Tracking Developers’ Conference, San Francisco—Sam Massih from Invensense described the various market segments and applications for wearable sensors. The latest wearable devices are becoming consumer friendly and moving to populations outside the hard-core fitness adherents.
Although originally designed for sports training and the elite athletes, the latest activity tracking gadgets are moving into the general fitness markets. Sports specific devices for running, biking, swimming, and other sports are offering sport-specific analytics and data tools to many people.
The big data aspects of the industry reflect the collection and analysis of large volumes of data from many users. The sensors supply data to a client device that uses an app to analyze and store data for performance evaluation. The client provides the access and services as well as any additional compute power to the sensor assembly.
The many facets of fitness, sports, and health and medical monitoring support the quantified self, a person who can use measured data to evaluate every day performance. The makers of the wearable sensors are capturing the various data we all generate in our activities and, through ease of use and access, can monetize those data.
Although accelerometers are the largest part of the market, the trend is towards activity tracking with multi-mode sensors. The minimum seems to be a full 6-axis inertial measurement unit. The industry has evolved from the accelerometer-based devices like pedometers, to activity tracking with multi-axis sensors plus barometric pressure to indicate altitude changes, to activity monitoring with full 6-axis combo sensors.
The next generation will include all of the active sensors plus wireless connectivity. Soon, the monitors will add biometric sensors to enable the addition of heart rate, body temperature, skin conductivity, and other body functions to allow the user to truly measure everything about the body functions.
Wearable sensors will also become mobile accessories. A smart watch could provide snapshots of incoming texts and e-mail while also keeping time. The most use app on smart phones is the clock, with an average person checking the time up to 150 times per day. A smart watch could reduce the power requirements to check the time and also provide the user with communications updates.
The fitness products are using accelerometers as the base sensor. Some can track session-based activity modes like the duration of a run or bicycle ride, but cannot distinguish the different activities autonomously. The devices extract features from the data to determine activity levels and duration, but a accelerometer can only generate about 6-8 feature sets from the data. Adding a gyroscope can provide greater accuracy while reducing code size for the analysis, since the combination can generate 20-25 features such as changes in activity as well as orientation information.
The combination of sensors provided greater accuracy than the accelerometer only products, and the greater set of features available reduces the size of the decision tree to determine specific activities. The system can keep the power demands manageable by reducing the duty cycle of the gyro to only be active under certain levels and types of acceleration are detected.
The combination sensors are supported with an SDK that includes sensor fusion tools and context-aware motion software. The SDK supports 9 sensors including acceleration, gyroscope, magnetometer, temperature, humidity, proximity, light, UV, and pressure. only the motion data are fused, and all other data are passed through raw. The fusion tools can be disabled if desired, and all sensor data can be accessed.
