Simpler than teleporting
Cisco is using AI and the IOT to take video conferencing to a new, almost Big Brotherish level
At the Cisco research & development labs in Oslo, Norway, the company has made artificial intelligence (AI) the core of its nextgeneration video collaboration tools by using technology found in self-driving cars.
The Nvidia chips used by Audi and Volvo that recognise people and objects and try to avoid collisions have been adapted for Cisco’s Webex products to focus on facial identification.
Cisco has also developed its own algorithms to identify faces and sounds. Extraneous background noise, like the tapping of keyboards, loud chewing, car alarms or dogs barking can then be automatically eliminated during video conferencing.
Future capabilities include extracting back-end data from a meeting, such as transcripts and translations of what was said.
Snorre Kjesbu, vice-president of Cisco’s collaboration endpoint technology group, believes the quality and quantity of data is central to holding “intelligent” meetings. “It’s about taking a relatively simple thing like telemetry, being able to get information back using the Internet of Things, and combining it with complex stuff like AI and machine learning,” says Kjesbu.
Cisco uses telemetry at its R&D facility in Oslo to track the presence and movement of people in meeting rooms and offices, so available rooms can be identified. Data protection laws in Europe allow a company to track numbers of people, but not to identify them individually.
Features that Cisco is working on include the use of intelligent sensors in a room that could register when someone enters or leaves, and start or stop video conferencing accordingly.
The technology could also register if using a smaller room would be more efficient.
John Restrick, chief technology officer for devices at Cisco, says: “We provide the hardware that goes into meeting rooms like cameras, microphones and touch panels, and along with our features that pull both training data and context of a business, we are able to leverage real-time data about what’s happening in a meeting.”
The aim is to make the technology “invisible”, such as a “one-button-to-push” system that eliminates the need to dial into a conference call and enter a PIN; or using “intelligent views” in video conferencing so that people who are speaking can automatically be framed, zoomed into and followed as they move.
Doing this without a person in the room to choose different cameras requires a lot of intelligence and work, says Restrick, “especially for complicated environments