Dr. Kloor observes that “since I was a small boy, people have been predicting that robots are the future, that our AI-driven robotic butler is just around the corner. Yet, decade after decade, I watched with disappointment as this prediction never became reality. Sure, over the past few decades we have seen companies like Toyota, Honda, and Boston Dynamics create walking spectacles, robots that can even dance and do parkour. But for all their antics, they can’t perform a single useful task and are really just billion-dollar vanity pieces with limited AI and limited practical use. They are a big disappointment to the public and to their investors. But does this mean we have been lied to? No, I don’t think so. It just means that the right company, with the right vision and the right technology, has not come along to tackle the problem. Until now. Beyond Imagination is that company, and we come up with the formula for commercial success, by focusing on not just building a humanoid robot, but an entire hardware and software platform to create market solutions.”

Dr. Kloor had the honor to work with Steven Jobs during his Next Computer years, and admired his brilliance at solving problems that people didn’t even know they had. Jobs also had the team and vision to create entire product ecosystems. He didn’t just make a smartphone; he created an operating system and app store that leveraged and amplified one another. Today, people can’t imagine life without their phones. In an analogous fashion, we built Beomni, a robotic humanoid with an AI brain that evolves via commercial human operation. We are also building a platform (operating system) that not only connects you to the robot, but will serve as a service portal, a marketplace for renting and purchasing robots, and an app store that will provide endless plugins. Like Apple, hardware and apps won’t just be from us, but from any developer. And our platform will serve as AirBnB for robots rentals and sales, not just from us, but from anyone.

Our go-to-market approach is like that of Tesla rather than Google’s Waymo. We are choosing to come to market rapidly by having our robot operated by human pilots, allowing the AI Brain to learn from that commercial use. Over time, this AI Brain will evolve and grow to fully assist the human, then become semi-autonomous, and finally become autonomous on a task by task basis. And autonomy for us is far easier than for a vehicle, as Tesla’s recent experiences demonstrate. 

Years ago, AT&T had this great slogan, “Reach out and touch someone.” The commercials illustrated how telecommunication enabled families to stay closer, that it was not just about talking but making a real human connection. But imagine if you could reach out and actually touch someone – if you could interact physically across any distance. You would be able to erase the limitations of locality. Instead of just talking and seeing people, you could interact physically with people, places, and things anywhere in the world. No longer would people be limited to the jobs, healthcare, and opportunities near them; the entire world opens up to them. With such ability, you can solve most of the world’s problems, because people are ultimately the solution to problems. There is enough technology and money in the world, but often not a person locally to make a difference. Work completely changes, as do healthcare and personal relationships. Voice and video have created trillions in revenue every year. Imagine the economic impact of physically interacting anywhere.

Our human brains have several lobes. At a gross level, each lobe has different functions. In a similar way, we are creating an AI brain, with AI components or sub-components that handle various functions like object recognition, audio, and natural speech recognition, path planning, motion, and haptic control. In addition, we have an experiential four-stage process built around machine learning combined with the concrete data gathered from physical operation of the robot or its digital twin. Over time, on a function-by-function basis, the AI grasps tasks like picking up and opening a bottle or stacking items. Unlike an autonomous car, Beomni doesn’t need to know how to do all tasks and can learn one by one. The closest a car can do is to perform the task of self-parking, which is hardly its primary purpose.

Automated robotic arms are commonplace in manufacturing, but humans still have to do the most complicated parts of the work, since some tasks are too complex or require adaptability. For automation, the robotic arm is simply going through a pre-planned motion path over and over again. If the holes that it expects to screw a bolt in are off by even a small amount, it will shut down or fail to hit the hole. Robots that are autonomous can recognize objects and sounds and adapt, so if the hole is 5 inches to the left they can find it and screw in the bolt. In fact, if the bolt is missing, they can go in search of it, find it and bring it back to finish the job. Of course, other senses can be built-in, including touch.