No doubt the proliferation of voice assistants has changed the landscape in which we live. Whether it’s Alexa or Siri or Google Assistant, we’ve come to accept voice as a big part of the evolving tech picture.
Not satisfied with the conversation being just about voice, Seattle-based Synapse thinks we could use an interface with more contextual awareness, to bring artificial intelligence closer to real human interaction.
“It’s kind of like talking to somebody on the other side of a wall. You have no additional context,” Jeff Hebert, VP of engineering at Synapse, said about voice assistants.
So Hebert and company want you to meet Gerard, a demonstration of what the company is capable of. And while the tuxedo-clad robot is not a real product, its abilities are, and they provide a glimpse of how a machine can learn about a physical environment and understand not just voice commands, but human gestures.
While a video does show Gerard in action, Hebert said the results could easily be produced from a point on a wall in a room that could perform 3D and spatial mapping.
“All of the constituent parts here in terms of spatial mapping and depth mapping and voice recognition, gesture recognition, they all kind of exist out there in products and things today,” Hebert said. “What we were trying to show is how putting them all together in this way is kind of a next step for what digital assistants and things will turn into.”
The expectation is that something like Gerard would start in the consumer space, turning on lights or TVs or music or other devices it the home, or helping find lost items as it does in the video. It would certainly reach other parts of the market, and Hebert said they’d be pretty surprised if you didn’t walk into a physician’s office years from now and find this type of assistant. And in an industrial workspace, the technology could help someone avoid having to look at a screen and help boost worker output and enhance safety.
And when it comes to privacy concerns, Hebert stressed that all of the video processing involved in interactions with Gerard is happening onboard, using “a pretty beefy Invidia processing engine.” No information is being pushed to the cloud.
Synapse, which was founded in 2002 and acquired by Cambridge Consultants in 2016, has about 160 employees, and has been working on Gerard for a little over a year.
The demo is expected to help bring brand awareness for the company while bringing potential opportunities to either build similar systems or systems that use the technology.
“We could work on autonomous cars because there’s elements of machine vision and depth mapping and edge processing,” Hebert said as one example.