Artificial intelligence is undeniably good at playing games, regardless of its larger societal and environmental impact. In 1997, IBM’s supercomputer Deep Blue famously beat chess grandmaster Garry Kasparov, and nearly two decades later, Google’s AlphaGo managed to defeat a human champion at the much more complex game Go. Since then, another AI seems to beat humans at incredibly complex video games like StarCraft or Dota 2 every few months. Impressive as these feats are, AI has mostly outperformed humans by making us play on its home turf, as in, simulated and structured environments. But now, a team of engineers at the Micro Air Vehicle Lab at the Delft University of Technology in the Netherlands has built a lightweight smart drone that’s fundamentally altering that dynamic.
The team entered its AI-powered quadcopter in a high-profile drone race in Abu Dhabi that pitted AI-piloted drones against human-piloted ones. The niche sport, known as first-person-view (FPV) drone racing, has pilots steer remote-controlled drones through a series of small gates at blistering speeds of 60 to 100 miles per hour. They control the drone by using goggles that give a direct feed into a camera attached to the drone’s front. From a human’s perspective, it’s as if they’re flying through the sky.
The team didn’t just fish the race. They took first place, besting several human FPV champions along the way. It marked the first time an AI-controlled drone beat a human pilot in an international racing competition. Another team from the University of Zurich built an AI-controlled drone that won a race against humans three years earlier, but only in a controlled lab setting. This more recent example showed that a simply trained machine, with minimal onboard sensors and hardware, could be inserted onto a non-controlled test track meant for human operators and still come out on top.
The team won this race in November 2025 but new details about the science behind this achievement are published today in the journal Science Robotics.
“Instead of giving the robot increasingly powerful sensors and computers, we wanted to find out how much intelligence we could extract from very limited resources,” Delft University aerospace engineer and study co-author Christophe De Wagter said in a statement. “Drone racing provided an extreme environment in which to test these ideas outside the laboratory. It pushed us to develop AI that was not only fast but also robust, and able to adapt when external reference measurements were unavailable.”
Drone racing: a sport built on human-machine connectivity
FPV drone racing isn’t for the faint of heart—or, more accurately, for those with a weak stomach. Footage of drones flipping and somersaulting at extreme speeds can make even spectators feel sick. Pilots fly using a pair of joysticks that control the drone’s throttle, roll, pitch, and yaw. That human-and-machine team has to fly through several small race gates that make up a lap. The drone finishing with the fastest lap times wins. It’s an extreme sport where even a microsecond of a mistake can lead to a violent crash.
Though it’s not quite as popular as it was at its peak in the late 2010s, the sport still draws a considerable online audience. That’s pretty impressive, for a sport that only became possible thanks to simultaneous advances in several technologies. Namely lightweight batteries, miniaturized cameras, and the goggles and wireless video links that let pilots see what their drones see all had to come together for this extreme form of racing.
The fact that drone racing requires interacting with machines already equipped with sensors makes it a particularly interesting test bed for robotics and AI researchers. In 2022, the University of Zurich outfitted a drone with an AI model appropriately called Swift. That AI system collected real-time data pulled from the drone camera to “see” what it was seeing and to locate gates and avoid obstacles. An onboard neural network then parsed through all that in real time, making split-second decisions on the best routes to fly. Swift beat champion human pilots, a major feat at the time, but it did so on an indoor track measuring roughly 82-by-82-feet that the team had designed and trained the system for. When a condition even as subtle as the lighting changed, Swift’s performance suffered as a result.
Inspired by those results, the Delft team set out to build an AI racing drone that could better adapt to its environment. Like Swift, their machine called MonoRace uses a powerful AI model that interprets a video stream captured by the drone’s camera in real time. MonoRace also draws on an onboard inertial measurement unit (IMU), a sensor that tracks the drone’s acceleration and rotation. This IMU pulls data from onboard sensors as well as the camera, and helps the AI system better understand where the drone is position-wise on the track.
Timelapse image of the autonomous MonoRace racing drone flying through the final racing gate at Abu Dhabi in 2025. Image: Micro Air Vehicles Lab/Delft University of Technology.
That understanding of place makes it more equipped to fly without crashing even in less familiar environments. It also improves the drone’s perception over a solely camera-based approach, which can struggle when camera data is less than stellar. Remember, all of this is happening very, very fast.
“Flying at very high speed with only one camera is challenging because the images are blurry, deformed, and the AI system has very little time to precisely determine where the drone is,” Delft University robotics researcher and study co-author Stavrow Bahnam said in a statement. “We developed a robust way of combining what the camera sees with the drone’s own motion information, allowing it to navigate the track at high speed without GPS.”
What sets MonoRace apart from complex AI systems running on larger vehicles or humanoid robots is how stripped down it is. The entire system has to run on a tiny computer on the drone, light enough not to slow the drone down and efficient enough not to introduce lag. And yet, it’s still quite powerful, with the AI issuing an estimated 500 commands every second. Rather than train the model on one specific track, they trained it to learn how to adapt using a reinforcement learning approach. After many attempts of trial and error in a simulated environment, MonoRace figured out the best ways to quickly solve navigation problems and then apply that in the real world.
“Instead of first planning a desired flight path and conventional commands, our neural network learns to directly control the motors,” study co-author and Delft University Aerospace Engineering PhD candidate Robin Ferede added in a statement.
Putting MonoRace to the test
The team put MonoRace to the real test last November at the Abu Dhabi Autonomous Racing League, a major competition where some of the best human pilots battle to out. During its flights, MonoRace topped speeds of roughly 62 miles per hour as it looped, dove, and zipped through gates. It managed to beat out every other AI racing team in the competition, and also boasted a better time than three former human world champion pilots.
According to the team, these results make MonoRace the first AI-piloted drone to beat human pilots in an international racing competition, and could have an impact on drone racing and physical AI more broadly. With self-driving vehicles and even some humanoid robots entering larger-scale production, there’s an increasingly high demand to build AI systems that better grasp how to move about the physical world without barreling into everything around it. A drone that can do that while performing acrobatics at more than 60 miles per hour suggests that engineers are, at minimum, heading in the right direction.
So far, that hasn’t resulted in larger robots suddenly superseding humans at sports. The fastest robot sprinters, though admittedly quick in short bursts, also tend to crash into walls and burst into flames.
While China and Russia rush to build a SpaceX competitor, the UK is taking another route by developing an alternative to satellite internet that relies mostly on aircraft. Its Advanced Research and Invention Agency (ARIA) is funding 18 projects aimed at building aircraft that will deliver wireless communications to areas in the UK and the rest of the world that are not served by terrestrial networks. According to The Register, the program, called Enduring Atmospheric Platforms, aims to have a unit with a 300-watt communications payload aloft and holding station for a week.
These units are called high-altitude pseudo-satellites (HAPS), with ARIA allocating $94 million over three and a half years for these projects. There are three technical areas in focus at the moment: TA1, which aims to develop the underlying technologies; TA2, which looks into the integration and testing of the developed systems; and TA3, which will concentrate on actual deployment.
Two startups have proposed multiple drone systems that will cover the entire United Kingdom — Carousel, by Menapia, wants to build a constellation that can hover at altitudes of up to 12 miles, while Enduring Drone Networks, by VFP Aerospace, proposes a relay of units that automatically fly back to their ground station for charging and servicing, with a cost of about $67,000 per drone.
Two other proposals suggest wireless power — one is Scalable Laser Ltd, which is working on a laser-based technology that pairs high-power laser diodes on the ground with advanced photovoltaic receivers in the air. On the other hand, Space Solar Engineering Ltd wants to use radio frequencies, like microwave, to deliver power to airborne units. There’s also a more novel approach from the University of Bath, which plans to use atmospheric gravity waves — using the upward movement of air caused by weather fronts and geographical features and the subsequent pull of gravity that brings it back down — to extend the range and endurance of HAPS aircraft.
"This research expands the UK's leadership in high-altitude platforms to become the place where this industry is designed, built, and scaled," says ARIA program director Rico Chandra. "Getting there means backing bold ideas, from fixed-wing solar aircraft to designs nobody predicted, including aircraft that fly on spinning wings instead of propellers."
Half a year after Iranian drone strikes knocked out multiple Amazon data centers, the US tech company has acknowledged the permanent loss of some customer data that was hosted in Bahrain and the United Arab Emirates.
Amazon Web Services was “unable to restore access to the resources and data” hosted in some of the war-damaged data centers, according to an AWS dashboard update posted on September 15. The development was first reported by Reuters and suggests that the Iranian strikes inflicted catastrophic damage on the data centers.
Customer data was irretrievably lost in one of three AWS availability zones in the United Arab Emirates region, specifically the mec1-az2 availability zone. Each availability zone is serviced by one or more Amazon data centers.
Half a year after Iranian drone strikes knocked out multiple Amazon data centers, the US tech company has acknowledged the permanent loss of some customer data that was hosted in Bahrain and the United Arab Emirates.
Amazon Web Services was “unable to restore access to the resources and data” hosted in some of the war-damaged data centers, according to an AWS dashboard update posted on September 15. The development was first reported by Reuters and suggests that the Iranian strikes inflicted catastrophic damage on the data centers.
Customer data was irretrievably lost in one of three AWS availability zones in the United Arab Emirates region, specifically the mec1-az2 availability zone. Each availability zone is serviced by one or more Amazon data centers.
It's time for TechRadar's annual tech awards, and this year we're doing it differently. For the first time, we have a complete set of categories where the winners are chosen 100% by you, the TechRadar readers.
The TechRadar Readers' Choice Awards 2026 comprises nearly 100 categories, but we know that not everyone is interested in voting in every one, so on this page you can vote in just our Phones and Cameras categories.
These include tablets, ereaders, drones and more, as well as various phones and cameras awards. You can vote in the survey just below — or follow the links to see other specific categories, or to vote on the full list instead.
The shortlists were chosen by TechRadar's editors based on the best products we've tested or the most popular products of the year. You'll need to vote before 00:01am PST / 03:01am EST / 08:01am BST on Wednesday October 7th 2026 for it to be counted.
Once the deadline has passed at 00:01am PST / 03:01am EST / 08:01am BST on Wednesday October 7th 2026, we'll tally your votes to find the winners in each category.
We'll announce the winners in our big TechRadar Awards 2026 Week, starting Monday October 19th. We'll reveal the winners of the Reader's Choice Awards, as well as the winners of our TechRadar Product of the Year Awards, which are chosen solely by our expert judges.
How many products will do the double? We'll be curious to see — but right now, all you can do is vote for what you think should win, so vote away!
Recently, operators from the Department of Active Operations and the Department of Unmanned Systems of the Main Intelligence Directorate of the Ministry of Defence of Ukraine carried out a joint deep-strike operation with the Ukrainian Defence Forces, during which another oil refinery on Russian territory was struck. The refinery in question is ‘Slavyansk-ECO’, located in […]
The Ukrainian Defense Forces struck the Volgogradpromproekt facility in the Russian city of Volgograd during the night of September 11. Ukraine used FP-5 Flamingo cruise missiles in the attack. This was reported by Fire Point on its Telegram channel. The Flamingo cruise missiles successfully reached their intended target. The Dnipro OSINT community reported on Telegram […]
Chinese-made drones are flying out the bodies of flood victims and delivering food to survivors in the aftermath of devastating flash floods on the border of Nepal and China’s Tibetan region.
Nepal’s army has been using four DJI FlyCart 100 drones donated from China to make 10 to 16 aid deliveries each day, according to Reuters. This comes after a partial glacier collapse in the Himalayan mountains unleashed a sudden surge of water, mud, and debris that swept through border crossings, valley communities, and hydropower projects on August 26.
The DJI Flycart 100 Drones come with a 98-foot rope and winch system for picking up and dropping off heavy loads. The website of Chinese drone company DJI describes the drones as being able to carry between 187 and 220 pounds, depending on the battery configuration.
Chinese-made drones are flying out the bodies of flood victims and delivering food to survivors in the aftermath of devastating flash floods on the border of Nepal and China’s Tibetan region.
Nepal’s army has been using four DJI FlyCart 100 drones donated from China to make 10 to 16 aid deliveries each day, according to Reuters. This comes after a partial glacier collapse in the Himalayan mountains unleashed a sudden surge of water, mud, and debris that swept through border crossings, valley communities, and hydropower projects on August 26.
The DJI Flycart 100 Drones come with a 98-foot rope and winch system for picking up and dropping off heavy loads. The website of Chinese drone company DJI describes the drones as being able to carry between 187 and 220 pounds, depending on the battery configuration.
Satellite imagery and new data reveal the aftermath of the recent strike on Novorossiysk. As a reminder, during the night of September 9, Ukraine’s Defense Forces – including the Security Service of Ukraine (SBU), Unmanned Systems Forces, Ukrainian Navy, the GUR Defense Intelligence active operations department, and the State Border Guard Service – attacked a […]
Many have tried and failed to produce drones to rival DJI's, but the Potensic Atom 3 is a DJI drone alternative that's actually worth considering — and not just on paper, but in actual flight conditions.
The folding quadcopter impresses with level 5 wind resistance and a 10% bigger battery than its predecessor, the Atom 2, making for robust flight performance even in tricky conditions. It's gargantuan RC PTD 2 touchscreen controller (in the pricier bundle) feels top-drawer, too.
Its 3-axis gimbal camera isn't half-bad, either — being equipped with a 1/1.3-inch sensor and offering 4K video up to 60fps in standard, HDR or Log profiles, slow-motion recording, and up to 50MP stills in RAW and JPEG.
The larger sensor, higher 4K video frame rates and bitrates, plus improved battery life and transmission range are all meaningful upgrades on the Atom 2, and make the Atom 3 a compelling DJI Mini rival, especially given Potensic's competitive pricing.
Despite a few handling issues (which I'll get into below), the Atom 3 could have pinched top spot in our beginner drones guide... six months ago. The bad news for Potensic is that the DJI Lito X1 is now in the picture, raising the bar for any would-be challengers.
For practically the same money, the Lito X1 is the slightly more polished drone overall — and it features object sensing, which the Atom 3 lacks, delivering this incredibly useful feature at a new-low price point (compared to DJI's Mini 5 Pro). Ultimately, the Lito X1 gives consumers so much for the money that the fight here feels unfair.
Would I still recommend the Atom 3? Absolutely. It's way better than the Ruko U11Mini 4K, which we recently tested, and the next best thing for consumers who don't want, or can't buy, a DJI drone (US pricing is displayed on the Potensic website, suggesting it could become available there, unlike new DJI products, but currently Potensic is locked out of the US too). Ultimately, Potensic's best drone yet is a highly capable beginner option — but it remains one big step behind the DJI competition.
The Atom 3 feels very much like a DJI Mini drone (Image credit: Tim Coleman)
Available now direct from Potensic and via Amazon; not available in Australia at launch
The Potensic Atom 3 is available now directly through Potensic's website and via Amazon in the UK. There's US pricing on the Potensic site, but the company told me the Atom 3 is not available there, while it's yet to hit stores in Australia.
Pricing varies considerably depending on the region and the bundle. In the UK, the base package costs £339.99, with the price rising to £579.99 for the Fly More Combo with RC PTD 2 bundle (which I tested), which includes two additional batteries, a charging hub, case, and the touchscreen RC PTD 2 controller in place of the standard smartphone-based remote.
US pricing starts at $429.99, while the Fly More Combo, which includes the standard smartphone-based remote, costs $549.99. At the time of writing only two of the five bundles have pricing — none of which include bundles with the touchscreen RC PTD 2 controller. However, despite pricing being on the Potensic site, the drone is not available there "due to FCC regulations", Potensic told me, and that "our team is actively working on the relevant certification to make it available in the future".
At the time of writing, the Potensic Atom 3 is not available in Australia, but Potensic does have a dedicated Australia website which is the best place to check for future availability.
DJI's Lito X1 is the Atom 3's obvious rival, and it costs from £359 at Amazon UK, so just a little extra money gets you that drone's massively useful object-sensing feature. However, like other DJI drones, the Lito X1 is not available to buy in the US.
Price score: 4/5
(Image credit: Tim Coleman)
Potensic Atom 3: Specs
Potensic Atom 3 specs
Potensic Atom 3
Camera:
12.5MP 1/1.3-inch CMOS sensor (with 50MP mode)
Video:
4K 60fps
Flight modes:
3x speeds, 7x Quickshots, 3x AI tracking
Storage:
MicroSD card slot, no internal storage
Battery:
2,590mAh / up to 40 minutes flight time
Charger type:
USB-C / Battery Charging Hub (Fly More Combo)
Weight:
249g / 8.8oz approx (C0 EU class)
Dimensions:
148 x 91 x 62mm / 5.83 x 3.58 x 2.44 inches (folded)
Potensic Atom 3: Design and build quality
Folding mini design, weighing under 249g
USB-C charging and MicroSD card slot, but no internal storage
Optional touchscreen RC PTD 2 controller is huge
The Atom 3's hard-plastic design is strikingly similar to recent models in DJI's Mini and Lito series of sub-250g folding quadcopter drones. It's a familiar formula — the drone folds down to a coat pocket-able size and, like most beginner models, weighs under 249g. Sub-250g drones are exempt from the strictest drone registration requirements in most regions, including the UK and US.
At the rear of the drone is a USB-C slot for charging the device directly, and for transferring images from the inserted microSD card. The card slot is also at the rear — there's no internal storage.
I was able to transfer files directly onto a MacBook Pro via that USB-C wired connection, but when I tried connecting the memory card directly my computer was unable to read it. Another option for file transfer is to download the desired files through the Potensic app, which is a reliable albeit slower process.
The Fly More Combo with RC PTD 2 bundle — the priciest of all the bundlesTim ColemanThe drone with its protective coverTim ColemanTim ColemanTim ColemanTim Coleman
A clip-on cover ably protects the camera and gimbal mechanism when the drone isn't in use, and this is locked in place with a rear clip by stretching an elasticated band that connects the two (see above).
The Fly More Combo (with RC PTD 2 remote controller), which Potensic supplied for this review, comes with a padded zipped case that holds the drone, the charging hub with its two additional batteries, and the sizable controller. The case is notably larger than the one for the equivalent DJI Mini 4K bundle.
The touchscreen controller feels high-quality; the screen is huge, detailed, and provides a 1080p feed up to 60fps, with Potensic's latest PixSync 5.0 video transmission system boosting the range from 10km to up to 16km, and reducing latency to 150ms. In short, the on-screen feed is smooth and immersive beyond distances that you'll ever realistically fly over.
I found certain movements of the controller's joysticks a little jerky, which spoiled some of my footage, but with practice I was able to find the sweet spots of those controls for smoother moves.
The USB-C and microSD card slots at the rear of the Atom 3Tim ColemanThe 3-axis gimbal cameraTim ColemanPotensic supplied me the dedicated ND filter kit. However, the lens guard in the press sample I received was glued in place, so I couldn't use these filters. Potensic assured me that retail units will be fine. Tim ColemanThe RC PTD 2 remote controllerTim ColemanTim Coleman
Design score: 4.5/5
Potensic Atom 3: Features and flight performance
Battery life quoted at up to 40 minutes; the drone typically lasted around 20 minutes in real-world testing
No object sensing, just the usual downward-facing sensors for landing and takeoff
Decent selection of flight and tracking moves, but some have frustrating restrictions
Potensic quotes a battery life of 40 minutes for the Atom 3, which is an improvement over the Atom 2, thanks to a 10% (approx) higher-capacity 2,590mAh unit, and on paper, this performance beats the comparable DJI; however, I didn't get close to this in real-world use.
The best battery performance I got during testing was more like 25 minutes on a calm day, while in windy coastal conditions I rarely, if ever, got more than 20 minutes. The drone's level 5 wind-resistancerating (19-24mph) seems fair, though — I used the Atom 3 in such conditions, and I found flight to be impressively stable and smooth, even if headwinds further drained the battery as I fought to bring the drone home when power was low.
When battery life drops to 10%, the drone automatically initiates landing (avoiding it simply dropping from the sky), so that's something to bear in mind, especially if you're flying over risky areas such as bodies of water.
The Fly More Combo includes three batteries and a charging hubTim ColemanBattery capacity has been upped to 2,590mAhTim Coleman
The Atom 3 has downward-facing sensors for takeoff and landing, with the useful Return to Home feature, but there are no front-, side- or rear-facing sensors, like you get with pricier drones equipped with omni-directional object sensing.
Six months ago I wouldn't have expected any more from a drone at this price point, but DJI has since brought object sensing to the much more affordable Lito X1. If Potensic is to go up a level, adding this incredibly useful feature is the logical next step for the Atom series.
Return to Home (RTH) worked flawlessly, with the drone automatically updating its position at takeoff — there's also an on-screen guide indicating where the RTH position is in relation to where the drone is in the air. It's possible to update the RTH position mid-flight, say if you move to a different location during flight and want it to land at your new location.
Downward sensors assist takeoff and landing, but there are no other sensors for obstacle avoidanceTim ColemanTim Coleman
For regular flight, there are three speeds to choose from, with the Sport Mode being the quickest at 16m/s (horizontal), 5m/s ascent and 4m/s descent. It's not the quickest drone around, but it's zippy enough for beginner use.
I really did miss object sensing, given that I regularly fly in object-packed areas to capture dynamic footage, so I needed to be extra cautious about where I flew the drone. It is possible to get smooth and stable-looking footage, though, after much practice to get a feel for the controller's joysticks.
There's a good selection of DJI-style 'QuickShot' flight moves, which perform an automated move using a selected subject as the reference point. To do this, you drag a box over the subject using the controller's touchscreen (or your smartphone's screen if using the basic controller). There are seven automated moves in all, including pull-away, rocket and boomerang.
I was happy to see three tracking modes too: Follow, Parallel and Spotlight, each of which keep your subject in the center of the frame as they move. 'Follow' tracks your moving subject from behind (though remember there's no object sensing), while 'Parallel' does the same but from the side. 'Spotlight' simply keeps the subject central as they move around the frame.
The tracking modes are hamstrung by several frustrating limitations, while the omission of object sensing further limits where you can fly. For example, to initiate Follow, the angle of the camera needs to be pointed down a certain degree — in other words the drone needs to be quite high up.
For me, the most dynamic 'follow' footage would be from a lower angle. It's possible to override the limit in the app's menu, which requires checking the option to say that propellor guards are installed (whether in reality they are or are not). Even with the options selected to free up the drone for tracking from a lower altitude, I regularly found that the drone reset itself, reimposing the restriction.
I also found that the tracking functions gave up altogether in less-open areas, when the GNSS (GPS/ GLOSNASS / Galileo / BeiDou) signal was weak. One example was trying to get tracking video of myself biking in my local woods — the drone kept stopping just a second or so into sequences. I had no such issues in the same scenario when testing the DJI Mini 5 Pro (which also has omni-directional object sensing).
Those flight moves are great to have, but adding object sensing, and addressing the other limitations, would make them way more useful for dynamic footage.
Flight performance score: 3.5/5
Potensic Atom 3: Image and video quality
1/1.3-inch CMOS sensor, 4K video up to 60fps
Wider 24mm f/2.8 lens (vs 26mm f/1.8 lens of the Atom 2)
RAW stills and flat P-Log video color profile provide editing latitude
Video and photo quality are an upgrade from the older Atom 2, with the Atom 3's camera using a larger 1/1.3-inch CMOS sensor that's comparable to those in the DJI Mini 4 Pro / Lito X1, rather than the Atom 2's 1/2-inch sensor. There's also a slight change to the lens — it's a wider 24mm f/2.8, where the Atom 2 uses a 26mm lens with brighter f/1.8 aperture, which somewhat counters the sensor-based advantage the Atom 3 has.
4K video recording up to 60fps can be done the easy way or the hands-on way, using the standard or detail-rich HDR profiles for ready-to-go footage, or the flat P-Log profile, which holds more detail in highlight and shadow areas, but which requires editing to taste later.
I tried all three profiles at 30fps and 60fps, and then took P-log footage to Adobe Premiere Pro for editing and compared the footage. There are incremental image quality gains when using P-Log, but on the whole I found myself happy with the quality of HDR footage, saving myself the editing hassle.
The 24mm f/2.8 cameraTim ColemanSame scene as before, but with the 2x digital zoom employedTim ColemanThe exposure here was too brightTim ColemanWhereas manually adjusting exposure using exposure compensation kept the highlight detailsTim ColemanTim ColemanThe minimum focus distance doesn't allow for selfies!Tim ColemanTim Coleman
In addition to 4K video, there's slow-motion capture up to 7x, but only in 1080p HD resolution. Meanwhile, stills are reasonable: 12.5MP as standard, or 50MP in the '8K' mode, with JPEG and RAW formats available.
Where photo quality impresses is in how much tonal depth is retained, although images benefit from a little sharpening afterwards. Another option for further increasing shadow and highlight detail is the 'bracket' mode, which combines multiple photos at different brightness values into one with a wider range of tones, but the standard mode already does a good job.
Interval and burst shooting modes are also included, while a panorama mode combines multiple images into one ultra-wide perspective that covers 180 degrees, with the drone automatically capturing multiple frames one after the other.
I'd say image quality is just about on par with the DJI Lito X1 / Mini 4 Pro, and it's clearly a cut above the recent Ruko model.
Tim ColemanTim ColemanTim ColemanTim ColemanTim ColemanTim ColemanTim ColemanTim ColemanTim Coleman
Image and video quality score: 4.5/5
Potensic Atom 3: Testing scorecard
Potensic Atom 3
Attributes
Notes
Rating
Price
Competitively priced, but matched by the overall better DJI Lito X1.
4/5
Design
DJI-like build quality, immersive touchscreen controller (optional), no internal storage
4.5/5
Features and flight
Stable flying, sticky controls, no obstacle avoidance, flight modes hit with limitations
3.5/5
Image and video quality
Smooth video even in windy conditions, decent 4K video quality, punchy HDR setting, and detailed stills. Digital-only zoom, though
4.5/5
Should you buy the Potensic Atom 3?
Buy it if...
You're a US buyer and beginner pilot With the Lito 1 and Lito X1 unavailable in the US, the Potensic Atom 3 gives budget-conscious American buyers another option to weigh against older DJI models like the Mini 4K.
You want attractive footage straight out of camera, plus the option of flexible editing Attractive standard and HDR footage looks great, with no editing required, while a flat P-Log profile better enables editors to get the look they like from footage with wide dynamic range.View Deal
Don't buy it if...
You plan to fly in places with lots of hazards The Atom 3 has no obstacle-avoidance safety features whatsoever, but the similarly priced Lito X1 does. Some of its flight modes, such as Follow, didn't work for me in obstruction-filled areas such as forests, either.
You want a super-cheap 4K drone The Atom 3 is the best DJI alternative for beginners, but the Atom 2 remains a decent buy for less, if you can live without the latest model's upgrades.
You want precise joystick control Even the pricier RC PTD 2 controller by Potensic lacks the finesse I've experienced with DJI's alternative controller, making cinematic movements a challenge. View Deal
Potensic Atom 3: also consider
DJI Lito X1
DJI's Lito X1 is not available in the US, but for other regions it's the better pick over the Atom 3, for a similar price. Video and photo specs are similar, but the X1's flight modes are more polished and flight features are superior, including obstacle avoidance, which is lacking in the Atom 3. DJI has really made it tough for others to compete with the X1.
The Atom 3 is a solid upgrade over the Atom 2, with a larger sensor, high video frame-rates and 10-bit color depth with P-Log color profile. However, the Atom 2 offers a similar flying experience, and costs less.
I tested the Atom 3 in a range of conditions — in daylight and at dusk, and in both calm and windy weather
I shot video at every available resolution and frame rate combination, and still photos in all modes
I assessed battery life and wind resistance over several flights in various conditions
Potensic sent me the Atom 3 Fly More Combo kit with the RC PTD 2 touchscreen controller. The kit included the drone, battery hub with three batteries, controller, ND filter pack, spare propellors and a USB-C to USB-C charging cable.
I conducted multiple flights in daylight and in low light at dawn / dusk, and in both calm and particularly windy coastal conditions. I tried the various flight settings and modes, and regularly ran the three batteries down, making notes of real flight times and flight stability.
I took the same video sequences in the various color modes; standard, HDR and P-Log, and at 30fps, 60fps and 6x slow motion, and compared the 4K footage. I also tried the various photo modes, including RAW & JPEG capture, panorama and 8K (48MP) modes. I edited photos in Adobe Camera Raw and videos in Premiere Pro.
Depois de décadas de testes e protótipos falhados, o Exército norte-americano vai receber 26 lasers anti-drone de 30 kW, num contrato de 465 milhões de dólares com a AeroVironment. Este é um marco que pode acelerar toda a indústria de armas de energia dirigida.
At the MSPO international exhibition in Poland, the Ukrainian unmanned systems developer UKRSPECSYSTEMS unveiled an attack drone with an X-shaped wing for the first time. It is reported that the RAM-2X system has already proven itself to the military, but had previously remained classified due to security restrictions. To demonstrate the operation and effectiveness of […]
Well, tell that to president Trump. On Thursday, he slapped a 100 percent tariff — he sure loves round numbers and levying imports — on larger foreign-made drones.
Speaking to the New York Times, Scott Mlakar, chief of the Lake County Drone Unit outside Cleveland, called the tariffs financially onerous and lamented that it would prevent many communities from buying them, adding that his unit uses them to hunt crime suspects at night and search for lost children.
“Most of this, in our humble opinion, is lobbying by US manufacturers to eliminate foreign competition,” he said. “This is not national security. It’s protectionism.”
The tariff is explicitly designed to target popular Chinese drones, while making exceptions or imposing much lighter tariffs on drones made from allies like Japan, Taiwan, South Korea, and the EU. In particular, these steep duties will apply to drones that use thermal imaging or weigh more than 55 pounds. In addition, the Federal Communications Commission is weighing reclassifying some drone technologies like lidar as “military grade,” Ars noted, which would restrict who can access these drones even further.
Drones have a lot of other practical uses beyond law enforcement, obviously, including monitoring wild fires and assisting rescue operations. Amid heightened backlash against Flock, though, they’ve come to be seen as another surveillance apparatus.
But weaning the US off Chinese drone tech is supposed to be good for national security, according to the Trump administration. One small problem: a single Chinese company, DJI, makes more than 70 percent of the world’s commercial drones, according to the NYT. Good luck replacing all those.
Plenty of other cops are bellyaching about the tariffs, too, extolling how irreplaceable that awesome Chinese made drones have become to their operation.
Ars spotted a Texas police officer named Jason Lee, for instance, who complained to the FCC that “there is currently no economically viable domestic replacement” for DJI drones. “More than 80 percent of state and local law-enforcement drone programs surveyed in 2024 used DJI aircraft,” he added.
“The fact that a domestic product exists does not necessarily mean it is a viable substitute,” Lee said. “A replacement must be operationally comparable, economically obtainable, supported by an adequate supply chain, and accepted by pilots who must rely on it during real-world incidents.”
What began as a routine police response to a reported break-in turned into a brief standoff between nature and technology. Last week, law enforcement in the city of Kirkland, Washington (just northeast of Seattle) responded to a 911 call from a resident reporting loud banging inside their home. Officers arrived expecting a human intruder, but were instead greeted by a more literal cat burglar: a trapped and confused bobcat.
Not wanting to risk a claw to the face, the officers backed out and reassessed the situation. They ultimately decided to deploy a remote-controlled drone into the house, to monitor the feline intruder from a safe distance. Video captured from the drone’s perspective and released on Facebook, shows the frustrated bobcat standing by a staircase, growling and arching its back at the hovering machine. After a brief standoff with the buzzing, metal alien, the bobcat carefully retreated, sidestepping into an adjacent empty room.
Police were then able to close their door and keep the bobcat isolated until a team from the Washington Department of Fish and Wildlife arrived. Wildlife officials then sedated the cat, loaded it into a crate, and transported it to nearby Forbes Creek. The feline was then released back into the wild, startled but safe. As for how a bobcat ended up in the house in the first place, Kirkland Police believe the adventurous intruder climbed a tree and jumped through an open window, possibly lured in by the resident’s indoor cat. Both the owner and its likely shocked cat were unharmed, according to the police. Thankfully, curiosity didn’t kill this cat.
The Kirkland Police Department did not immediately respond to Popular Science’s request for comment.
The aerial deployment was made possible by a public safety drone program the Kirkland Police began in 2022. Today, they reportedly maintain five drones and have 17 FAA-licensed pilots ready to operate them. The typical use cases for those drones include search and rescue, tactical search, scene documentation, and, apparently, bobcat removal.
The bobcat before its release near Forbes Creek. Image: Kirkland Police Department.
Drones are changing the way police surveil wildlife…and people
Drones and robotic dog-like quadrupeds are reshaping how local communities approach wildlife management. In Japan, authorities deploy a combination of camera-equipped drones and robotic Monster Wolves to locate and deter bears following a record spike in human-bear encounters driven by urban sprawl and declining farmland. In the United States, police in Santa Fe, New Mexico, recently used an aerial drone to track an errant bear wandering near a school. Expanded drone programs are also playing a greater role in locating lost or missing pets. Just last month, firefighters in t Florida used a remotely controlled drone to find an overly adventurous dog trapped in a storm drain.
But this increased prevalence of drones comes with other, more controversial concerns as well. Police forces around the country are ramping up “drones as first responder” programs, which civil liberty critics say could introduce unreasonable surveillance. Law enforcement around the country has already used these drones to monitor political protesters. Meanwhile, Immigration and Customs Enforcement (ICE), is reportedly interested in using Boston Dynamics’ quadruped robot “Spot” to aid in immigration operations. Private security companies like Flock and Axon are also investing heavily in aerial surveillance drones.
This week, the US Army used a high-energy multipurpose laser to shoot down three drones near the US-Mexico border, which official reports claimed were “posing a physical threat to US military personnel and CBP partners,” referring to Customs and Border Patrol.
The operation took place between the night of August 25 and the early morning of August 26 in the vicinity of the Rio Grande Valley in Texas and was led by the Joint Task Force Southern Border (JTF-SB), which operates under US Northern Command. Authorities confirmed that, during the operation, agents used technology derived from the Army’s High-Energy Multipurpose Laser (AMP-HEL) system, a military program focused on implementing high-energy lasers for new defense capabilities.
Available information suggests that this initiative began in September 2025, when AeroVironment, a company specializing in defense technology, delivered the first two functional prototypes of these laser-based weapons to the US Armed Forces. Two new, improved systems were delivered as the program was expanded in December of the same year. According to the manufacturer, this particular platform, called the Locust Laser Weapon System, is designed to be mounted on military vehicles, as the Army has done with the versions it has received.
This week, Romania accused Russia of sending a drone boat loaded with explosives to a vital European natural gas platform in the Black Sea. A Romanian F-16 fighter jet used its cannons to disable the surface drone so that explosive ordnance experts could blow it up safely.
On August 20, the Romanian Coast Guard first spotted the maritime drone a few hundred meters away from the Neptun Deep project’s main drilling platform, where several hundred workers were busy with the platform’s ongoing installation, Defense Minister Radu Miruță said in a Facebook post. The offshore gas project would make Romania the largest gas producer in the European Union once it starts full production in 2027.
To protect the lives of the workers and the offshore gas infrastructure, the Romanian military scrambled two F-16 fighter jets to stop the drone with cannon fire, Miruță said. Army specialists later carried out a controlled detonation, as seen in a Facebook video shared by the defense minister.
Amazon wants to grow its Prime Air delivery drone service sixfold to reach nearly 500 cities and towns across the United States by the end of 2026. The planned expansions from Amazon and its rivals could make the convenience of delivery drone services available to tens of millions more people in the coming months—but will also test more communities’ tolerance for increased noise from daily drone flights.
On August 19, Amazon announced it would soon launch Prime Air in the metropolitan areas of Chicago; Cleveland; Atlanta; Syracuse, New York; and Boise, Idaho, with more to come later in the year. Each Prime Air site offers delivery drone service within an area of about 175 square miles, with delivery times ranging from half an hour to an hour.
Prime Air already operates out of Amazon warehouses in 10 cities across seven states. Those include primarily suburban neighborhoods of Phoenix; Tampa; Kansas City; Omaha, Nebraska; Baton Rouge, Louisiana; Detroit; Houston; and San Antonio, Dallas, and Waco, Texas.
Login
