Low power, inertial, pedestrian navigation IC complements GPS

Low power, inertial, pedestrian navigation IC complements GPS

Feature articles |
PNI Sensor has announced SENtrace, the first coprocessor for wearables providing accurate and ultra-low power pedestrian tracking indoors, as well as in urban canyons - this is good news especially near tall buildings and especially in large cities like NY city where I have much personal experience with this phenomenon; so, anywhere that the global positioning systems (GPS) signal goes missing or is inadequate.
By eeNews Europe


I spoke to George Hsu, PNI Sensor Chairman and resident tech guru, and he commented that SENtrace is a small custom ASIC that uses about 10x lower power than GPS demands. By building on PNI Sensor’s proprietary embedded algorithms, it enables existing ultra-low power inertial sensors to track users when there is little or no GPS signal available.

Another feature that I really like is that it greatly reduces overall battery consumption because it overrides and deactivates power-hungry GPS when it’s not needed. In typical configurations, the IC uses about one-tenth the power of GPS for computing each location point.

The chip provides tracking to one-metre accuracy over 100 metres travelled, supplying step-by-step data in lieu of extrapolations between two location points. The PNI solution greatly augments GPS accuracy or even allows the designer to choose when and how to turn off GPS with three options: Based on time, distance travelled, or quality score.

PNI further explains; [this] inertial tracking technology’s performance is based upon distance travelled . It is typically 1% to 2% of distance travelled in normal use cases. So the error would grow to 10 to 20 metres for 1 kilometre travelled and so forth. Since this type of error is different in nature to GPS [that has an estimable error that is dependent on available constellation configuration, for any given fix] the two technologies are quite complementary to one another. For short distances Sentrace can provide better resolution in terms of positioning than GPS can as well as save power. Over long distances, GPS can be used to reset the inertial accuracy in order to bound the errors.

“We foresee a range of applications for SENtrace in wearables,” added Becky Oh, PNI Sensor President and CEO. “They include wrist-worn devices for locating lost children or elders and enhanced activity wristbands and smartwatches for athletes and fitness enthusiasts.”

I like the fact that the PNI solution has constant calibration to keep the system accurate.

The IC physical board footprint is 1.6 x 1.6 x 0.5 mm, and it is a 32-bit processor with a custom floating point unit (FPU) embedded with PNI’s sensor fusion tracking algorithms. These algorithms offload the tracking task from a device’s main processor, thereby saving battery power as well as not loading down the main computing power of the system.

Here is a simple block diagram of a typical system in which SENtrace is a central feature.

SENtrace’s algorithms work with any sensor manufacturer’s sensors, allowing wearables original equipment manufacturers (OEMs) to select from a wide array of commercially available accelerometers, gyros and magnetic sensors. This has been one of PNI’s signature aids to designers who can choose from a wide variety of supporting components for their system.

Specifications comprise;

Operating Specifications

Accuracy; 2° RMS

Data Update Rate; 400 Hz max.

Latency; < 1 video frame

Outputs; Heading, Pitch and Roll; Rotation Matrix; Quaternions; Sensor Data

Supply Voltage (VDD); 1.6 – 3.3 VDC

Current Consumption; Peak: 800 µA, Typical: < 100 µA at 7 Hz Kalman update rate; Standby 5 µA

I²C Interface Frequency; 100 – 3400 kHz

Operating Temperature; −40 – 85 °C

SENtral’s 9-axis sensor fusion technology is built on a tailor-made ultra low-power coprocessor from EM Microelectronic, Swiss specialists in producing ultra-low power microelectronic components.

PNI Sensor; www.pnicorp.com/products/sentral-sensor-fusion/

Linked Articles
eeNews Embedded