Editor’s Note: This article first appeared on Colin Angle’s LinkedIn page and was republished with permission. To stay updated with his blogs, follow Angle on LinkedIn and Twitter.
For an eternity, experts have equated robot intelligence with autonomy. They were wrong.
For the world to move past the precipice and into the golden age of robotics itself, the robots of today will need to move from being autonomous to truly being intelligent.
What does this mean?
There are three steps that build upon the baseline of autonomy that we need to take.
The first thing we have to do feels a bit counterintuitive: we have to make robots more responsive. This feels a bit in conflict with the idea of autonomy, but in reality is actually quite complimentary. How so? While a good robot should go about its learned routine autonomously, it should also be perpetually ready to take instruction from its owner to meet a current need that its routine would not expect, or necessarily know immediately how to handle. Consumers expect that robots will deliver a personalized experience that seamlessly fit into to their lives. To do this, they need to listen and take direction.
Let’s think of an example specific to Roomba: Let’s say you’re the proud parent of a 6 month old baby (congratulations!). When baby eats, baby is in a highchair, and when he or she is in a highchair eating, you better believe a mess will appear on the floor beneath the chair. Once the mess is made … you don’t want to wait for Roomba to start its regularly scheduled routine to eventually clean under the highchair. You want to be able to ask Roomba to come clean under the highchair right away.
Related: iRobot wants to make it easier to clean up after pets
If Roomba encounters something on the floor with which it’s not entirely familiar – say a wet spill of an unknown liquid (go ahead and let your mind run wild) – wouldn’t it be better for Roomba to ask its owner what the owner would like Roomba to do before attempting to drive over and vacuum it?
This is what I mean when I talk about robots that are responsive, providing a personalized experience. To put it simply: they need to be able to take direction and ask for it when circumstances require. Beyond improving performance and outcomes, a robot that is responsive drives the relationship between robot and owner closer to one based on trust. And as someone who runs a company that has sold 25 million robots, I can assure you trust is not granted at the point of purchase. Trust is earned over time. It’s why so many owners of Roombas watch them vacuum when they first bring them home. It must be proven to them that Roomba will do what it promises to do.
A robot must be collaborative. To accomplish this we must broaden the awareness and understanding of the robot beyond its immediate environment to people and other robots as well. For example, if Roomba encounters some unknown liquid substance on the floor, perhaps it asks its owner for guidance… but also offers the option for a Braava robot to come mop up the wet spill instead. Should the owner of the robot provide guidance that Braava do the job, Roomba could then signal to Braava to come do the job, and provide it location information on the spill to be cleaned. Braava then comes and does the job, and Roomba goes out completing the rest of its mission. Collaboration complete. Over time, one could expect the robots to take the proper collaborative action without asking the owners for input as they learn what liquid spills they should and should not clean.
This collaborative dynamic also further evolves the relationship between robot and owner from one of trust to that of a trusted partnership. Why? Because a partner not only listens, a partner learns. Given enough inputs by the owner, a robot should be able to learn, and when the unusual circumstance crops up again in the future, take action without asking for input as the robot has learned what to do from past experience.
The third step is that robots must take towards true intelligence is to act as part of a larger system. Think of your house as a potential system. It has rooms, and we tend to go into rooms for a task and leave for something else. If your Roomba understood what devices are in what rooms, suddenly your house could operate more like a system, for example, allowing you start watching a TV show in one room, and have that show follow you into another room, which makes for an optimal experience for you. In addition, the room you were in previously would know to turn itself off – lights off, TV off – thus allowing you to save energy.
There’s one more important dynamic that is critical – the most critical – to robots stepping into the golden age: people must trust them as partners. It’s why it’s so critical that robots deliver progressively on the three steps I listed above. By delivering on these steps robots will engender the kind of the trust that is at the foundation of a true partnership. With trust granted by people and partnership achieved through that trust then and only then can robots be given permission by people to empower people to do more.
At iRobot, we’re not just talking about these steps, we’re taking them. You can already ask your Roomba robots to clean a specific room in your house on demand via voice command to Amazon or Google smart speakers or via the iRobot HOME app. We call it Directed Room Cleaning and it’s our robots’ most popular digital feature – even more so than scheduling! Roomba and Braava can already collaborate together to clean an entire floor of the house or a specific room depending on your preference through Imprint Link technology. With respect to building a system for the home – the smart home – Roomba already holds the keys to the proper foundation for one: spatial understanding.
The movies promise that autonomy is the answer, but it is a false promise. The golden age of robotics won’t dawn until robots truly understand your home, can listen and take direction. Once this is accomplished, a world of possibilities exist. Robots talking to robots. Robots providing intelligence and instructing other devices in your home. There’s much more to come on that front, but for now I’ll just ask this: if I told you the smart home of tomorrow is actually a robot … would you believe me?
Stay tuned because I’ll explain why in a future blog post. Until then, let me know what you think it’s going to take for us to step into the Golden Age of Robotics.
About the Author
Colin Angle, a co-founder of iRobot, has served as chairman of the board since October 2008, as chief executive officer since June 1997, and prior to that, as our president since November 1992. He has served as a director since October 1992. As a co-founder and chief executive officer, Colin provides a critical contribution to the board of directors with his detailed knowledge of iRobot, its employees, client base, prospects, the strategic marketplace and competitors.
Colin previously worked at the National Aeronautical and Space Administration’s Jet Propulsion Laboratory where he participated in the design of the behavior controlled rovers that led to Sojourner exploring Mars in 1997. He holds a B.S. in Electrical Engineering and an M.S. in Computer Science, both from MIT.
Many of us have pets that don’t do what they are supposed to do outdoors (or in the litterbox) all the time. It would be great if my Roomba stopped when it encountered liquid on the floor, and pinged my cell phone app to ask what to do next? In this instance, the Roomba has now become a spill detector in a basic away (warn of indoor flooding one way?). Same with unknown intruders that it detects during working hours. Why limit to floor-sweeping? Why not basic house patrol with vision/motion/heat sensors? I would pay $X per month for that security service rather than pay ADT… Lot’s of recurring revenue models that can be built on the iRobot ecosystem in the future.