By Dacia Rivers
Tamagotchi pets were a big hit with kids in the late 90s. Created in Japan, these tiny, handheld gadgets contained computer images of various animals, which users could feed and care for all the way from hatching to old age. Brad Knox, CEO of Emoters, is taking the Tamagotchi concept to the next level by creating small robots designed to live on a table or desk.
“What we’re trying to do with this first generation product is turn the dial on illusion of life as far to the max as we can,” Knox says. “The more we can achieve that, the more it should be entertaining and fascinating, but also become compelling for a much longer term than other entertainment electronic characters that are out there.”
Users will be able to interact with these charming desktop pets in a variety of ways. Like live pets, they’ll respond in their own limited ways when their owners speak to them, and they’ll react to changing environments. On their own, Emoters will run around on their desks, interacting with other objects Knox develops for them, such as food, treats, toys or even “dirt” the robot will clean up for its owner.
Knox is also developing the robot hardware in a way that will allow users to purchase multiple units affordably and watch them interact with each other. “You can end up having a pack or herd of robots,” Knox says. “There’s an observational impact to see them playing games with each other and interacting.”
Anyone who owns a Roomba is familiar with the way the robot vacuums tend to fail at returning their own charging devices when their work is done, winding up stuck under a sofa, drained of battery. To avoid this problem, Knox is working to ensure his Emoters will reliably return to their charging docks unassisted so they can be active for weeks, without any user effort or intervention.
Knox plans on launching a crowdfunding campaign for Emoters in early 2017. To prepare, he and his team have been working with a product design firm to get the robot hardware ready.
To develop the AI the Emoters will use, Knox has been using a technique known as Wizard of Oz prototyping. In this instance, he’ll use a secret puppeteer to control an Emoter to see how a person interacts with the robot.
“You get to test out what happens if the robot makes a whimpering sound… how do people respond?” Knox says. “Do they interpret it the way we think they will?”
The AI Knox plans to use in developing Emoters comes from the academic research he conducted when he was at the MIT media lab. The Emoters team keeps logs of the secret puppeteer’s interactions, then uses machine learning to develop a model that determines the probability of the puppeteer’s reactions to certain interactions. Once they’ve built the model, they can give it the reins and see where it goes.
“This method of using machine learning to emulate a human improvising puppeteer reaches a level of organic behavior that really isn’t achieved through current methods of creating electronic characters sold as toys or non-player characters in video games,” Knox says. “In that process, teams of developers determine what a character would do in certain situations, but psychologists know that if you want to know what someone would do in a situation, it’s much better to observe than ask them.”
Knox believes the initial market for Emoters will be young professionals, especially those who tend to be early adopters of all things tech, along with makers, STEM students and people who are interested in robotics. Following that, he plans on expanding sales to children and teenagers.
Right now Emoters is pre-investment and pre-revenue. Working with a team of three part-timers, Knox is focused on getting the crowdfunding campaign kicked off to bring attention and financing to his product.
Being on the cutting edge of robot-human interaction, Knox has a wide range of thoughts on the future of AI and some of the perceived threats that can come along with sentient mechanics.
“I heard the head of one major robotics company say that companion robots should never show negative feelings, and for me that raised a red flag,” Knox says. “What if a kid is mean to a robot and it acts like nothing is wrong — what does that teach the child?”
Knox brings an undergraduate degree in psychology to his robot business, and he is eager to watch and learn as his robots interact with humans and vice versa. His goal? To always consider how interactions with robots could affect humans’ interactions with each other, and how he can make sure that effect is positive.