AI Emerging Company

Exploring Human-computer Interaction Startup Rokid’s Jungle

Four years after its founding, human-computer interaction and robotics company Rokid held its first major press conference on June 26 in Hangzhou. Dubbed “Rokid Jungle,” Founder and CEO Misa Zhu quipped that the presser was “a combination of Google IO and Apple WWDC.”

Four years after its founding, human-computer interaction and robotics company Rokid held its first major press conference on June 26 in Hangzhou. Dubbed “Rokid Jungle,” Founder and CEO Misa Zhu quipped that the presser was “a combination of Google IO and Apple WWDC.”

As Rokid’s goal is to produce smart products in the verticals and an open-source platform for the AI horizontal market, it will have to navigate a virtual jungle of project and platform research, development, production, sales, and services. At January’s CES 2018 Rokid debuted its open-source Rokid Full Stack Open Platform — a turnkey solution that provides third-party developers access to Rokid’s algorithm and hardware solutions and services.

At Rokid Jungle the company unveiled three new hardware products: KAMINO18, Rokid Me, and Rokid Glass.

Kamino 18 is a coin-size, voice-dedicated AI chip that reduces device power consumption by 30 to 50 percent, with an integrated neuronal processing unit, digital signal processor, and a digital-to-analog converter. With the latest built-in algorithms, Kamino 18 is applicable for smart speakers and smart children’s toys.

image (36).png
According to Zhu, “The biggest difference between the Kamino 18 and other AI chips is that it provides a full speech solution instead of simply using voice signal processing.” Rokid’s chip is also 30 percent cheaper than general-purpose chips on the market.

Rokid Me is a portable smart speaker equipped with the Rokid voice assistant and a six-microphone array for all-direction voice pickup within a 5-10 metre distance. The 4000 mAh battery allows eight hours of continuous use. This speaker has built-in voiceprint recognition technology, and users can simply shake or flip the device for music control. Rokid Me also supports Bluetooth calling and gameplay. Priced at US$120 (RMB 799) and available in four colors, Rokid Me will be officially released on July 26th.

The Rokid Glass prototype was first introduced at 2018 CES. The newly designed AR headset has shrunk 40 percent in size, weighs only 120g, and has a metal frame for better temperature cooling. Battery life is six hours and up to 60 hours on standby. Equipped with a Qualcomm Snapdragon 835 processor, built-in WiFi, a 13-megapixel camera with optical image stabilization, inertial sensors, voice guidance system and microphone array, Rokid Glass can be used for face recognition, object recognition, indoor navigation and speech recognition; with applications in social networks, public security and retail.

The AR headset is slated for mass production by the end of this year.

image (37).png
Misa Zhu left Alibaba to found Rokid in 2014. The company specializes in human-computer interaction research and AI software and hardware development. Rokid has raised US$148 million in four rounds of financing, including a US$100 million Round B extension led by Temasek Holdings earlier this year.

Zhu defines Rokid’s R&D focus as a tri-stage development from Home AI to Portable AI, and finally to Personal AI.

The first stage of the strategy however suffered a setback. Rokid Alien was a cool, first generation, feature-rich home AI robot — but it was priced at US$788 (5280 RMB). This proved too high for the market and mass production was discontinued. Alien is now a collector’s limited edition.

The company pivoted to the household hardware sector and launched the smart speaker Rokid Pebble. Equipped with the company’s self-developed voice-dedicated AI chip, Rokid Pebble had success at the more affordable price of US$256 (1399 RMB), and has since recorded some 100 million user-system interactions.

Rokid is on the same developmental highway as the other top funded AI speech startups in China. AISpeech (US$121M), Unisound (US$211.3M), and Mobvoi (US$252.7M) are also extending their business to hardware AI chips for speech recognition — e.g., Mobvoi A1 voice chip and UniOne — after spending the last few years in speech recognition research.

The increasingly competitive AI speech market is now virtually impossible for smaller players to enter, and it is crucial for companies to gain momentum and build up their own ecosystems to survive the competition. Says Zhu, “Rokid is confident with our ‘full-stack’ approach.”

Zhu sees Rokid Me and Rokid Glass as new generation flagship products for portable AI, which he is confident will be a human-computer interaction portal to the new AI era.

Source: Synced China https://www.jiqizhixin.com/articles/2018-06-28-5


Localization: Tingting Cao | Editor: Meghan Han, Michael Sarazen

0 comments on “Exploring Human-computer Interaction Startup Rokid’s Jungle

Leave a Reply

Your email address will not be published.

%d bloggers like this: