“Moxie has the heart of Mother Teresa and the brain of a thousand Albert Einsteins. We have encoded that into something that can express empathy, be super supportive and knowledgeable about any topic that you want to learn about,” said Paolo Pirjanian, founder of Embodied Inc and inventor of Moxie Robot, in an exclusive interview with AIM. 

With over 30 years of experience in robotics, Pirjanian has been extremely passionate about the field. After completing his schooling and PhD in robotics, Pirjanian worked at NASA on rovers for the Mars exploration project. Having gathered vast experience in the field, he moved onto entrepreneurship, building startups in robotics for over 20 years. 

The path has obviously not been easy with sceptics at every corner, however, the advancements and adoption in the field have only been promising. “I’ve been doing this for a long time and there’s no more exciting time than now,” he said. “Moxie is literally science fiction brought to life. Even 10 years ago, Moxie was completely a dream.” 

Moxie – A Child’s Confidante

These were the words former NASA JPL roboticist and CTO of iRobot, chose to describe the visually appealing, talking, child-like robot that weighs less than 4kg and stands tall at about 40cm.

Moxie’s necessity arises from the therapy gap for children with special needs. “Unfortunately we don’t have enough trained therapists to be able to address the problem. For every therapist in the US, there are 350 children that need therapy,” said Pirjanian. As a result of that skewed ratio, the cost skyrockets and the waiting list goes up. 

The need to address the problem becomes even more critical in case of autistic children. “With autism, the two things we know are that the earlier you intervene and the higher the frequency, the better is the potential outcome,” said Pirjanian. With higher frequency of traditional therapy sessions, the cost is exorbitant, ultimately depriving children of these sessions.

Apart from helping children with neurodevelopmental challenges, Moxie also serves as a help for children dealing with traumatic experiences who are unable to express themselves. “If not [addressed] it becomes a scar in their mental health that can manifest itself in different ways in the future, maybe 40 years down the road or 10 years down the road,” he said. 

While, therapy and social emotional development is one aspect of Moxie, the friendly robot also assists with academic development. “Children can interact with Moxie, and ask as many questions as they want. Moxie is going to be super patient and infinitely knowledgeable thanks to generative AI. Large language models have practically included the entire human knowledge.”

Multimodal Moxie

“Moxie is way beyond sort of text-to-text LLM. It’s not just text in and text out, and this is an important distinction because about 90% of communication is non-verbal. So only-text contains about 10% of the content of interaction, and 90% is body language, intonation of voice, facial expressions, eye contact, nodding and emotional expression,” said Pirjanian. 

Moxie is a multimodal robot where the multimodal in and out mapping is at a very high level. The tech infrastructure involves a number of models including computer vision, voice recognition, sentiment analysis and others. 

In addition to collaborating with AI model providers, the company has proprietary models too. “We have a partnership with OpenAI, so we use their large language models as part of this. We use off-the-shelf automatic speech recognition technology too, but everything else is custom developed by us in the past eight years.”

Hallucinations being a common problem with LLMs has also been carefully mitigated with these robots as children are their audience and the robot should not push conversations to inappropriate corners. 

“One of the things we have done over the years is collect a lot of data and train a model that moderates the inputs and outputs of the large language models, and then gently nudge the conversation into what parents would consider appropriate, or sensitive to the situation, so that it doesn’t allow the child to go off the rails,” said Pirjanian. 

The company releases a new version of the software that powers Moxie every month, where the quality assurance team constantly tests and monitors everything that Moxie does.

Source: MoxieRobot

Giant Leap in Robotics

Having worked in the field of robotics for decades, Pirjanian expressed his enthusiasm on the strides of developments in robotics. In the past, merely possessing a computer vision algorithm capable of recognizing obstacles was considered a daunting task. 

“The good news is that just in the last 10 years, we have seen massive advancements. Things that we have been working on for decades and not really making progress are being solved left and right.” Pijanian believes that we are at a ‘watershed moment’ where robotics is finally becoming real and will impact every aspect of one’s life. 

“It was extremely hard to do that with a ton of computation, a lot of data, and many complex algorithms. Now, a high school kid can figure out how to do that with tools that we have built in AI, computer vision, and robotics,” he said. 

Foundation Model for Robots

Pirjanian’s company Embodied Inc, gels well with the current advancements in robotics, where 2024 is the year of embodied AI. With massive advancements in robotics, especially humanoid robots, Pirjanian believes ChatGPT also had a role to play.

“We are at a moment where we saw the power of ChatGPT last year,” said Pirjanian. “I think that’s a completely game-changing discovery. These large language models will impact robotics.” He believes that foundation models for robotics understand the relationship between perception and action to perform a task. 

At the recent NVIDIA GTC, in addition to showcasing how the company is powering every major humanoid robotics company, CEO Jensen Huang unveiled GR00T, a new foundation model for robots. Interestingly, a robotics startup Physical Intelligence is also working on building foundation models that can control any robot for any application. 

Eyes India

Pirjanian visited Bengaluru last month to understand the Indian market and work towards building Moxie, catering to the Indian audience. “Indian society is very diverse. One of the reasons I’m here is trying to figure out partnerships, figuring out the right strategy for how to launch here, and what languages we need to support,” said Pirjanian, hinting at the vast number of official and unofficial languages in India. 

“It’s likely that we may launch with Moxie speaking English and then gradually start introducing support for other languages as we better understand the market here.” Pirjanian confirmed that Moxie would enter the Indian market by 2025. 

The roboticist was also positive of the robotics developments in India, witnessing a lot of interest from investors here. “It’s the very beginning right now, but I think with the infusion of capital and talent, it will start spurring new spin-offs, companies and start-ups that will build the future of AI and robotics in India.”