I never imagined that consumer and industrial drone maker DJI would one day consider bringing a robot vacuum cleaner to the dining table, but I was wrong. And when you think about it, why wouldn’t they? After all, the company’s incredible squadron of drones is capable of tracking users across difficult terrain, even in forests where there are tree branches everywhere, with breathtaking precision and speed.
Since navigation and obstacle avoidance are prerequisites for any successful robot vacuum cleaner, it makes perfect sense to use established drone navigation technology in a robot vacuum cleaner, or even in a robot lawn mower. And since DJI would probably never dream of sharing its navigation technology with anyone, you might as well make your own robot vacuum cleaner. And that’s what happened.
I’ve already given the new DJI Romo P Transparent a very favorable review after putting it through its paces and seeing it work for a week. I also spent a lot of time analyzing its cleaning power on hard floors and carpets. However, one facet of this newcomer really surprised me and that was his very good ability to avoid obstacles.
Article continues below
In fact, I think the DJI Romo P’s obstacle avoidance system represents one of the most technologically sophisticated navigation systems ever deployed in a consumer robot vacuum. Unlike simple, cheap robot vacuums that rely on bumper switches or basic infrared sensors, the Romo P combines two fisheye vision sensors with three solid-state LiDAR units that DJI says bring “millimeter-level obstacle detection” to the home.
When combined, this multi-sensor data allows the Romo P to identify not only large, basic obstacles like furniture legs and walls, but also extremely thin obstacles, as narrow as 2mm thick. And that means it’s also capable of avoiding thin telephone cables of any length, a myriad of toys including Lego bricks and, if approached head-on, even flat objects like train tickets and playing cards. This level of discrimination is quite important, because thin objects from a credit card are among the hardest for robot vacuums to detect.
Navigation and obstacle detection hardware itself only collects data, but it is equally important to make intelligent decisions from that data. Therefore, DJI and, it must be said, other big players like Roborock and Dreame, are also integrating built-in machine learning algorithms, including AI-assisted learning, into their premium robots that interpret inputs from the combined sensors. These algorithms allow the robot to recognize different types of obstacles and choose the most appropriate course of action.
Sometimes it squeezes past narrow furniture legs to clean nearby, while other times it leaves plenty of room for most smaller items. This contextual knowledge apparently helps minimize cleaning interruptions and prevents the robot from getting stuck or inadvertently spreading messes, which I observed when the Romo P actively avoided a large dollop of tomato ketchup during my review’s cleaning test.
I also noticed the logical routes the Romo P takes each time it performs a task, and I can only assume that this is also due to its spatial awareness and ability to perform almost human-like movements through a maze of obstacles.
For my revision test, I tried to prove this and set up an obstacle course using a variety of objects, large and small. I was frankly amazed at how quickly the Romo P navigated around obstacles without stopping for breath or spilling anything. In fact, it was like watching an Olympic slalom skier in action.
Anyway, it got me thinking. So for this article I put together two more tests, even more extreme and not only focused on the Romo P, but called on my trusty Roborock Saros 10R, another robot vacuum with very decent obstacle avoidance.
By the way, the Romo P offers three levels of obstacle avoidance in the DJI Home app: Obstacle Avoidance Priority, Standard, and Cleaning Priority. For the following tests, I selected Standard mode because it better balances the gap between cleaning and avoiding.
Obstacle tests Roborock Saros 10R vs DJI Romo P
I’ve been using the Roborock Saros 10R since April 2025 and, along with its stablemate the Saros 10, it’s one of the best robots I’ve ever had the pleasure of reviewing. So I thought I’d see how well it would fare against DJI’s new Romo P when it came to negotiating a tough obstacle course.
For this comparison test, I carefully positioned a wide range of obstacles, including dog toys of different sizes, telephone cables of different lengths, three bottles of titchy diffuser oil, a small leaflet of stamps, a fake train ticket, a packet of cigarette paper, two thin black lids, a silver key, a small screwdriver and a sock.
I created two different tests because I wanted to see if either bot learned from previous mistakes. However, I can’t say categorically if they learned anything from test 1, because I’ve never seen a robot do the exact same thing twice. They are robots, after all, and they literally have minds of their own.
Anyway, back to the two tests which started with the Saros 10R followed by the Romo P. Both models were instructed to leave their dock position (the Saros in the hall and the Romo in the living room) and head towards the kitchen on a patterned rug. They were then ordered to return to the dock.
Even though they both came from different regions, I made sure that both robots were given equally difficult tasks, or as close to each other as possible. Having both robots start from the same area would have involved remaping the entire house to one of the robots and I simply didn’t have the time or, for that matter, the inclination. However, as the video shows, both robots faced equally complex assault courses.
Trial 1
Roborock Saros 10R
The Saros 10R happily entered the room and simply headed straight for the cigarette papers and a telephone cable, but then managed to reach its goal without causing chaos in about a minute. His return journey was more haphazard and he rode over a small cable and the stamps and, after thinking carefully with his camera in front of the cigarette papers, he decided, what the hell, and continued on his way, carrying the packet into the lobby before dropping it off as he approached his base station. His return trip lasted 35 seconds.
DJI Romo P
Now it was the Romo P’s turn, and for this I selected its “Standard” obstacle avoidance mode which is the best option for better cleaning performance. He immediately turned to the key and, rather disappointingly, walked over it. Typical! He also looked through the book of stamps before continuing with a strange detour around a small telephone cable. However, he still completed the task in just 48 seconds.
His journey home was also a bit finicky and lasted exactly a minute, although he took one look at the little screwdriver (understandable) and rolled over the stamps. But this time, he stopped at the end of a key before taking a detour through a rumpled hair before returning to the platform.
I have to say that they both performed extremely well in this test and I was particularly surprised by how the Saros 10R handled the course. However, the Romo P was clearly much faster at accomplishing this task.
Test 2
Roborock Saros 10R
The Saros 10R approached the packet of cigarette paper and pushed it aside before heading towards the black telephone cable. He then carefully made his way between the two diffuser bottles before noticing he had rolled over the short cable and backing away. He then continued by peeking at one small black plastic cap before crushing the second, completing the task in 45 seconds.
His return trip was less successful. He immediately stepped over the black cork and, after thinking about the small bottles for some time, ran headlong onto the black cable, which he dragged most of the way home. Time: 38 seconds
DJI Romo P
A perfect race for the gold medal. This time, the Romo avoided everything, including the key, the pack of cigarette papers, the stamps and the two black caps. An astonishing achievement in 48 seconds flat.
However, Rom(e)o was very arrogant on his 50-second trip home and rolled over the screwdriver, stamps and cigarette pack (twice) as if he had had enough tests for one day and just wanted to get home as quickly as possible. You have to laugh.
Trial 3
DJI Romo P
For this final test, I used only the Romo P and selected its “Obstacle Avoidance Priority” mode which further enhances its navigation capabilities for ultra-cluttered homes. I even added some smaller obstacles like a thin piece of cardboard, along with the black caps and stamp booklet from previous tests. Although the Romo P did a little more thinking and a few detours, it made it through the course with no major issues to report other than its side brush marring the thin cardboard slice. Another gold medal.
While these tests don’t represent the real world in any way (at least I hope so), they just show how refined the navigation is on these two models. Still, while the Saros 10R’s obstacle avoidance can be considered perfectly acceptable for everyday scenarios, I have to give the winning rosette to the Romo P for a more precise and faster approach to overall navigation and the perfected art of obstacle avoidance.
Follow TechRadar on Google News And add us as your favorite source to get our news, reviews and expert opinions in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on YouTube And TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp Also.
