Why your Roomba takes a strange path to keep your floors clean

Any robot vacuum cleaner will remove at least some dirt from your floors. However, its extent and behavior in rooms varies greatly from model to model.

The most important factor behind this is the navigation system of a robot. Its navigation technology, associated with software, determines the actions of a vacuum cleaner. This plays a huge role in how a robot cleans a space. As might be expected, some robotic vacuums perform the task better than others.

Read more: How to prepare your home so that your robot vacuum cleaner does not remain blocked.

Fortunately, our brand new Robotic Vacuum Test Room at our warehouse lab in Louisville, Kentucky, can help us show the differences between robot vacuum cleaners, including their perception, interaction, and movement in a robot. physical space.

Robotic navigation on a budget

There are three main types of systems typically used by robot vacuum cleaners to navigate a space. The first is a simple collection of collision sensors, wheels, brush and cliffs. They tell robots when they hit or are about to hit objects. With this information, they can slow down or change course. In addition, these sensors help vacuum cleaners avoid falls from stairs.

You tend to find these systems in robotic vacuums with limited budget. The advantage is that they are much cheaper than more complex machines. The SharkNinja Ion S87 at $ 349 and Eufy Robovac 11S at $ 199 are examples of products with this style of navigation.

The disadvantage is that they operate randomly, encountering obstacles and swirling around the room. The very first iRobot Roombas did the same. Unfortunately, the ground cover is incomplete. Stains in tight places (corners, table legs and chair) attract a lot of attention. Open areas, however, are probably aspirated once (or maybe not at all) as the robot moves in a straight line until it detects something in its path.

These machines also take a long time to operate, about three times longer than the most advanced robot vacuumers need to attack the same area.

Of course, long cleaning times matter little if you tend to vacuum when no one is home and you have all day to do it. When the company has to arrive in 45 minutes or in other limited situations in time, it is a problem.

Visual or optical navigation

Other robotic vacuums combine the basic range of collision sensors with a main visual sensor complemented by a lens. These voids use a navigation algorithm called VSLAM (or simultaneous localization and visual mapping). The optical system can identify cue points on the ceiling, as well as judge the distance between the walls.

VSLAM also calculates the relative position of the vacuum in a room in real time, which allows the bot to create a map during the cleanup. Vacuum robots that work in this way navigate in a room with greater efficiency, by systematically cleaning the floor in a logical pattern. They will not waste time vacuuming in certain areas of a room on which the robot knows that it has already traveled. As a result, they can cover the same area in less time and with better coverage than a robot based solely on physical sensors.

The current range of iRobot Roombas, such as the $ 1,099 i7 + and $ 1,299 S9 +, have this kind of navigation system. The same goes for high-end Ecovac models like the Deebot 711 at $ 499.

Vacuum robots with visual control have some disadvantages. Since their optical sensors need at least a certain amount of ambient light, they have trouble making their way into completely dark rooms. Compared to the basic models, you will also pay extra for these smarter robots.

Laser navigation

The lidar (light detection and telemetry) is another way used by vacuum robots to detect their environment. This is the same type of technology that you find in many prototypes of autonomous cars such as those of Waymo and Uber. All Neato Botvac use this method, including the $ 829 Botvac D7 Connected, the current flagship model of the company.

The high-level Ecovac Deebots like the $ 549 Ozmo 930 also have an integrated lidar. In this sophisticated system, a turret-based laser mounted on top of the vacuum robot illuminates the objects to help the robot determine their location and distance. Lidar vacuum cleaners can also detect the size and shape of objects in their path.

They are actively analyzing their environment too. This is why these machines tend to cover floors with extreme efficiency. For example, the Neato Botvac D7 and the Botvac D6 cleaned the floor of our test room in just under 21 minutes.

The SharkNinja Ion S87, valued at $ 349, and its basic navigation, spent an hour and nine minutes cleaning the floor of our test room. Similarly, two budget-limited Ecovac machines, the Deebot 500 at $ 279 and the Deebot 600 at $ 249, both had a cleaning time longer than 1 hour (60 and 64 minutes, respectively). The longest though was the RoboVac 11S Max at $ 199 (100 minutes, 34 seconds), also the cheapest model in this group.

A shorter runtime is not the only benefit for lidar. Associated with the SLAM algorithm (or localization and simultaneous mapping), these robots also create detailed maps on the fly. You can also perform useful interactions with these cards. For example, you can delete virtual limits or create restricted areas at will for the robot. These vacuums also navigate in the dark if necessary. All this is great. Remember that you will pay a premium for these machines. They usually occupy the highest echelons of the market.

Hybrid systems

A new approach adopted by some robot vacuum cleaners is to associate several navigation technologies in a system. This includes optical brush, cliff, wheel and optical sensors, as well as laser emitters. There are not many products that do this for the moment.

Today, you can buy the Electrolux Pure i9 for $ 499. This unique robot vacuum cleaner is equipped with a pair of front-ignition lasers. Sitting in the middle of them, on the front side of the vacuum cleaner, is also a large optical sensor behind a lens.

Even with all this technology, the movement of Pure i9 in our test room seemed confused. Like Neato and Roomba machines, she did not ride with confidence. Instead, it was lost in jerks, pivoting constantly in different directions.

With so many tools, as well as enhanced software and processing power, hybrid navigation robots can offer unmatched levels of automation and intelligence. I think the next Ecovacs Botvac 960 seems particularly promising. Ecovacs says the vacuum will be able to identify objects such as shoes, clothes and stacks of toys.

And the company says that AI-based recognition of the robot will learn new objects over time. Maybe this list will include damage from pets and other wet, loose debris or tights. This would be a welcome update, potentially preventing your floor and carpet from getting even more messy than before cleaning the robot vacuum cleaner.

A note on our new robot vacuum cleaner test room

We performed straight-line tests, based on cleaning performance, for robot vacuum cleaners, but this actually only gives part of the image of the success of a robot vacuum cleaner in your house. The quality of its use, its extent and duration are also important factors.

To help us collect this information, we have built an industry standard test room, as specified by the International Electrotechnical Commission (IEC), the international standardization body that governs, among other things, Vacuum testing of robots for manufacturers.

In our test room are objects and challenges designed to mimic what a robot will encounter when it cleans a room. This includes buildings designed to mimic large pieces of furniture such as sofas or dressers, smaller objects such as lamps or table and chair legs, and even surface irregularities such as carpets, transitions between a floor and furniture. electric cords.

A camera mounted above captures a bird's eye view of all the action. From there, we can determine the path taken by each vacuum cleaner during its cleaning cycle. This system also allows us to calculate the area covered by a machine and the time required to do it.

Look for other vacuum robot tests from us in the near future. At the moment, we can at least conclusively say that not all robot vacuum cleaners are the same and that the way a robot navigates in a room will not only affect its cleaning performance, but also the time it takes to clean. It is necessary to perform his work.