The SENtral-A2, the tiny and ultra-low power coprocessor, is designed to free designers from time-consuming algorithm creation for wearable and smartphone apps. PNI claims it offers the most expansive algorithm feature set and development framework of any product of its kind, offering designers a rapid means of providing more personalized and contextual experiences to consumer devices.
The SENtral-A2 coprocessor features such built-in functions such as FFT, filters and a context framework that speed development of context, activity and gesture algorithms. The algorithm context framework is designed to enable customer-specific feature extraction. SENtral-A2 runs the Google CTS test suite that ensues Google Android apps will run on a mobile device.
Standard algorithms embedded in SENtral-A2 include: 9-axis sensor fusion, game rotation vector, geomagnetic rotation vector, glance gesture, gravity, linear acceleration, pick up gesture, significant motion, step counter and detector, tilt detector and wake, shake and turnover gesture. Context algorithms include walk, jog, run, still, bike, vehicle and tilting.
Optional algorithms include activity monitoring such as push-up, sit-up, sleep states—waking, light, deep and heart rate monitoring. Device orientation and motion algorithms indicate how the device is held or carried, such as ear, handheld in front or at side, or in pocket, as well as auto detection of wrist-worn vs. not wrist worn. Finally, environmental algorithms include touch, proximity and light.
The 1.7mm x 1.7mm chip is pin-compatible with earlier SENtral versions and features more on-board memory for the addition of algorithms and to batch more data. The device routinely runs more than 140,000 floating point operations per second at ~200 µAmps average current consumption at 1.8V.
