AI-powered machines are revolutionizing transportation, taking over factories and managing our daily lives and home environments. Although most adults are now accustomed to smart devices, children aged 4–9 have thus far had less cause to communicate with virtual assistants or operate other AI-powered products. A Boston-based company is changing that with its new intelligent plush toy.
Meet “Woobo,” the cuddly children’s smart companion robot. Like Siri or Alexa, Woobo listens to and learns from its human owner. It can answer questions, engage in conversations, and even tell jokes. Children squeeze Woobo’s stubby limbs to activate its home menu, or press on its ears to activate voice commands. Woobo’s touchscreen meanwhile displays facial expressions and other visual elements such as videos and interactive games.
MIT alumnus Feng Tan is Woobo’s co-founder and CEO, and the main mind behind the smart toy. Tan got the idea for an educational children’s robot in 2015, when he was still a PhD candidate, and built a prototype during his one-month winter break. Despite looking something like a green mop with eyes, the first crude Woobo garnered both praise and the third-place prize of US$2,000 at the 2015 MIT Business Plan Competition. “I remember our product got the most applause and cheers,” says Tan.
Following the competition, a venture capitalist offered Tan seed funding. There’s good reason for such investment — according to P&S Market Research, the global personal robots market is projected to reach US$34.1 billion by 2022.
During the two years of Woobo’s development Tan interviewed over 200 families to gain a better understanding of what a children’s robot should look like and what it should do. He discovered for example that children prefer to manually activate Woobo rather than using the voice activation method associated with typical virtual assistants. Children also tended not to conduct multiple-round conversations, opting instead for single-round question-answer interactions.
Di Wang, co-founder & CTO of Woobo, built an intelligent system that returns an actual answer rather than listing documents or web search results. A PhD from Carnegie Mellon University, Wang won the TREC 2015 Live Question Answering Track in 2015 and 2016. The responses returned by Wang’s system, dubbed WaterlooClarke, have been favourably rated by users on online Q&A interface Yahoo Answers, and the technology has been integrated with Woobo.
Wang’s technical team continuously monitors children’s’ most asked questions to update the system, which covers child-specific topics as well as open-domain questions such as “What should I do if my kitten is pregnant?”
In recent months Wang has pushed the technology to the next level by adding a social chatbot with a dialogue personalization feature. Empowered by end-to-ending deep learning and transfer learning, Woobo is able to start a casual conversation for example in the style of Batman, simply by studying Batman scripts. “Woobo can become Batman, Lincoln, or anyone,” says Wang, who is now in the process of designing and developing a unique Woobo character profile.
In late 2017, the company sent the first 3,000 Woobos to families who had pre-ordered the product through Woobo’s kickstarter campaign.
Sarah Losie of Ohio says Woobo has quickly integrated into her four-year-old daughter Violet’s daily life. The smart doll’s sharp question-answering ability has engaged the child, who also loves the games and multimedia stories. Losie told Synced that Woobo recently took Violet on a virtual adventure to China, which led to mother and daughter discussing other countries and languages, the Great Wall, and viewing photos of historical interest.
“We’ve had a lot of great conversations start because of Woobo. My daughter has a blast playing with him and learning from him, and I can tell that this will be a toy that will grow with her,” says Losie.
Tan says Woobo’s goal is to create a supportive environment in which AI can educate, entertain and provide companionship: “I hope Woobo can make children feel more relaxed at home, just like being with a puppy or a kitten.”
Journalist: Tony Peng | Editor: Michael Sarazen