Competing Sensors Crave Human Contact

Competing Sensors Crave Human Contact

PARIS – While CMOS imagers and infrared are well established as sensors in smartphones, today’s $64 million question is whether they will make room for radar and ultrasound inside handsets and, by extension, IoT devices. And if so, what roles will they play?  

The Mobile World Congress last week in Barcelona showcased a new generation of sensors and sensor fusion technologies, reflecting the intensifying competition among tech companies to create a new human-machine interface (HMI) -- going beyond touch.

There is no such thing as one-size-fits-all sensor technology, but the product segments pursued by sensor suppliers are notably similar: smart speakers and IoT devices (including smartwatches/smartphones, AR/VR headsets).

The new HMI modalities they hope to enable are touchless, micro-gestures and presence detection.

One recent breakthrough in the sensor field is the emergence of remarkably miniaturized ultrasound transducers and radar chips. Cost for these could still be a big issue compared to infrared, for example. But the tiny radars and ultrasonic sensors are now small enough to fit inside consumer electronics devices, promising a whole new way for a user to interact with a smart device.

Infineon, for example, last week announced at the Mobile World Congress a partnership with XMOS. The German chip company promised that a combination of radar and microphone sensors from Infineon and audio processor from XMOS will enable “far field voice capture” and beamforming with radar target “presence detection.”

The CEO of a Berkeley, Calif.-based startup, Chirp Microsystems, came to Barcelona to pitch her company’s single-chip ultrasonic time-of-flight (ToF) sensor, for AR/VR and wearable devices, in addition to smartphones and, eventually, automotive.

Elliptic Labs (Oslo, Norway) showed off ultrasonic presence detection technology for smart speakers (i.e. Google home, Amazon Echo) and other home devices. Guenael Strutt, vice president of product development at Elliptic Labs, told us, “Presence detection is now the Holy Grail” for designers of home intelligent assistants. In place of CMOS cameras or infrared technologies, “Designers of smart speakers are looking for alternative sensor technology to let their device know that ‘no one is home,’” he explained.

While the three companies mentioned here are going after similar product segments, each is positioning its technology for slightly diverging roles.

Voice capture
On an extremely noisy show floor at the MWC, Andreas Urschitz, president of the Power Management & Multimarket division at Infineon, found a perfect place to demonstrate what today’s iPhones are sorely lacking:  the ability to separate the voice of particular person Siri needs to hear.

Every time Urschitz asked his iPhone “what time is it now” at Infineon’s booth, Siri said, “I am not sure what you said.”

We all know about a smart microphone that combines a number of individual microphones linked by beamforming. Presumably, this is where complex trigonometric functions happen, which enables the individual microphones to join into a highly directional beam, which can focus on the current speaker.

Now, Infineon is saying that the combination of Infineon’s radars and XMOS’ beamformer can go further, directing the microphone to “a specific object precisely, even with the object moving and indistinctive noise.”

“The system can distinguish your voice in a distance of 30 feet,” said Urschitz. Used in the demonstrator on display were the XMOS audio processor serving as a sensor hub controller, Infineon’s MEMS microphones (four microphones) which form the beam, and Infineon’s 60 GHz 2Tx/4Rx radar IC with accompanying antenna and a 70dB SNR microphone that helps to overcome these impediments.

The sensor fusion of Infineon's radar and MEMS microphone with audio processors from XMOS provides a new building block for voice recognition. (photo: EE Times)The sensor fusion of Infineon�s radar and MEMS microphone with audio processors from XMOS provides a new building block for voice recognition. (photo: EE Times)

Asked how much it would cost to add such a radar/microphones/sensor fusion processor configuration, Urschitz declined to comment. But he explained, “Every revolutionary technology – like Bluetooth and WiFi – has gone through this.” He is confident that this will come down to “a tipping point” where the cost of integration becomes reasonable.

Next page: Understanding micro-gestures


PreviousIntel, Spreadtrum Demo Brainchild
Next    Softbank to Sell 25% of ARM to Saudi-Backed Fund