Recent developments in AI and robotics have sparked an interest in creating and selling household robots that can complete a wide range of menial tasks.
According to Tesla CEO Elon Musk, the company is developing a humanoid robot that could be used to do things like prepare meals and provide assistance to the elderly. Amazon has been investing heavily in robotics technology through the Amazon Robotics program, and they recently acquired iRobot, a leading robotic vacuum manufacturer. Known for its high-powered vacuum cleaners, Dyson announced in May 2022 that it would establish the largest robotics center in the United Kingdom to focus on the creation of domestic robots.
The increasing demand for these robots may mean a delay in their commercial release. Commercial use of household robots is still in its infancy, but devices like smart thermostats and security systems have found widespread adoption in recent years.
As a robotics expert, I can attest to the fact that creating a robot for the home is a much more challenging task than creating a smart digital device or an industrial robot.
Household robots, in contrast to digital devices, require physical manipulation of objects to complete their tasks. They are responsible for transporting dishes, rearranging seating arrangements, and bringing dirty laundry to the laundry room. For these tasks, the robot must be able to pick up and carry fragile, soft, and occasionally heavy objects with complex shapes.
Modern artificial intelligence and machine learning algorithms scale up well in lab settings. Yet they frequently fail when confronted with objects from the real world. That's because it's tough to model and even harder to regulate how closely two people touch each other. While these are simple tasks for a human to complete, there are still significant technical hurdles that must be overcome before household robots can handle objects as deftly as a human.
Both controlling and sensing the object's environment are challenging for robots. Some pick-and-place robot manipulators, like those used in assembly lines, have only a simple gripper or other tools designed for one specific purpose, like picking up and transporting a specific component. Robots have a hard time with objects that have an unusual shape or are made of a flexible material because they lack the efficient force feedback that humans have. It is still difficult and expensive to develop a robot hand that can be used for a variety of tasks and has movable fingers.
It's also important to note that traditional robot manipulators perform most accurately when used with a stable platform, and that accuracy drops dramatically when used with platforms that move around, especially when operating on a variety of surfaces. Before widely capable household robots can make it to market, the robotics community needs to solve the open problem of coordinating locomotion and manipulation in a mobile robot.
Work in a factory or warehouse, for example, typically takes place in a clean, orderly setting where tasks are performed in a specific order. Engineers can program the robot's movements beforehand or use quick and easy methods like QR codes to guide the robot to its destination. But it's not uncommon to find things in the house strewn about in a disorganized fashion.
A home robot's environment is fraught with unpredictability. The robot needs to search through a large area and zero in on the desired item. In order to get to the item and finish the job, you'll usually have to move some other things out of the way or clear the area. This requires the robot to have an excellent perception system, efficient navigation skills, and powerful and accurate manipulation capability.
Users of robot vacuums, for instance, are aware that the best models still require manual clearing of small furniture and other floor obstacles like cables. Furthermore, the robot has to function while people and animals walk nearby, adding another layer of difficulty.
There are a lot of tasks around the house that humans may find easy, but robots simply can't handle. Industrial robots excel at routine tasks because their movements can be easily pre-set. In contrast, domestic chores frequently present their own challenges, making it necessary for the robot to make decisions and adjust its course on the fly in order to complete them successfully.
Imagine yourself in the kitchen or washing the dishes. In the course of a few minutes in the kitchen, you might find yourself holding a bottle of oil, an egg, a spatula, a sauté pan, a stove knob, a refrigerator door handle, and a sautéed onion. When washing a pan, it is common practice to use one hand to hold and maneuver the pan while using the other to scrub away any remaining bits of cooked-on food and soap.
Machine learning has been used to great effect in recent years to teach robots to make judicious choices when picking and placing objects. But even the most advanced learning algorithms would have a hard time getting robots to learn how to use every possible kitchen tool and appliance.
Additionally, many houses feature obstacles like stairways, tight corridors, and lofty bookcases. Today's mobile robots, which typically have wheels or four legs, are unable to access these areas. Still not widely deployed outside of research facilities, humanoid robots would function more naturally in settings analogous to those humans create and arrange for themselves.
Creating robots that are designed to do one specific job, like robot vacuum cleaners or kitchen robots, is one way to handle complex tasks. It's anticipated that numerous variants of such tools will be created in the not-too-distant future. However, I think it will be quite some time before we see consumer-grade robots for the home.