NTT Docomo and Tomy have created a true conversation-starter for Japanese kids
NTT Docomo, Japan’s leading mobile phone carrier, has tied up with domestic toymaker Tomy to jointly develop a conversation-capable “robot” for kids. OHaNAS (Organized Human interface and Network Artificial intelligence System – and also a play on the Japanese verb hanasu, to speak) connects to Docomo’s cloud-based natural-language dialogue platform via smartphone, using Bluetooth and a companion app. It’s capable of analyzing a wide variety of expressions and word patterns to determine nuance and meaning in a person’s speech.
In a release, Docomo states that OHaNAS will allow “children and adults [to] engage in highly natural voice-based exchanges in Japanese.” It can talk about such topics as “personal information, meal suggestions, music playback, fortune telling, weather information, [and] Q&As.”
So, apart from the cute sheep-meets-robot shape, what makes it better than chatting with Siri on your iPhone?
“Docomo’s proprietary cloud-based natural-language dialogue platform is open to a variety of third-party developers, so they can customize the platform as they like,” Aya Hokamura, a NTT Docomo spokesperson, tells Tech in Asia. “Using this platform, OHaNAS can recognize and react to a specific user’s profile for chats based on the user’s preferences.”
When a user chats with OHaNAS, it remembers mentions of specific hobbies and personal likes and dislikes. When it responds to the user, it can be programmed to mimic the speech patterns of famous Japanese characters and mascots. It can also identify and suggest information relevant to a conversation by searching the web for keywords. One example the firm pointed out is a user mentioning “doing laundry” triggering OHaNAS to do a quick search for weather reports in case of rain or strong wind (Japanese people tend to hang their laundry out to dry, rather than using dryers).
OHaNAS marks the first time that Docomo’s dialogue platform has been employed in a third-party device, but a simplified version is already being used on Docomo smartphones. Shabette Concier, a voice agent similar to Apple’s Siri, has been preinstalled on some of the firm’s handsets since 2012 (before Docomo embraced the iPhone). The current cloud-based natural-language dialogue platform that powers OHaNAS is based on Shabette Concier and has been in development since August of last year.
The connection to IoT
The firm is serious about putting its dialogue platform to use in a variety of IoT devices, so OHaNAS could be viewed as a friendly field test for Japanese consumers – who don’t appear to have many hang-ups when it comes to chatting with robots. It’s also a great opportunity for Docomo to collaborate with new corporate partners. Hokamura pointed out the firm’s voice agent for automobiles as just one iteration of the platform’s future:
Interactive voice-activated devices will become increasingly important in the IoT era, when a diverse array of things will be connected to the internet. An excellent example of this is Docomo’s voice agent for communicating with automobiles. The agent reacts to driving circumstances, such as sudden braking [and] driving destinations […] and learns from previous conversations. Someday, we expect drivers to communicate effectively with automobiles using [it].
Tomy, the company behind Tomica die-cast cars and a variety of Pokemon and Naruto toys, plans to sell OHaNAS in Japan this fall. It will retail for JPY 19,800 ($158) and will only support Japanese, though Hokamura left the possibility of additional language support open for future releases.
Musio, a robot from California-based AI software company AKA, appears to be the closest English-speaking competitor to OHaNAS. The most basic Musio, which uses machine learning and natural-language processing to engage in conversations with users, can be had for $99 on Indiegogo, where it has raised a total of $55,000 of its $50,000 goal with 23 days remaining. Musio, which was initially conceived as a tool for teaching English to kids in Asia, doesn’t require a connected smartphone, but runs on its own quad-core CPU.
This post was originally published on Tech in Asia.
Feature Image Credit: Tomy