In December, NASA took another small, incremental step towards autonomous surface rovers. In a demonstration, the Perseverance team used AI to generate the rover’s waypoints. Perseverance used the AI waypoints on two separate days, traveling a total of 456 meters without human control.

“This demonstration shows how far our capabilities have advanced and broadens how we will explore other worlds,” said NASA Administrator Jared Isaacman. “Autonomous technologies like this can help missions to operate more efficiently, respond to challenging terrain, and increase science return as distance from Earth grows. It’s a strong example of teams applying new technology carefully and responsibly in real operations.”

Mars is a long way away, and there’s about a 25-minute delay for a round trip signal between Earth and Mars. That means that one way or another, rovers are on their own for short periods of time.

The delay shapes the route-planning process. Rover drivers here on Earth examine images and elevation data and program a series of waypoints, which usually don’t exceed 100 meters apart. The driving plan is sent to NASA’s Deep Space Network (DSN), which transmits it to one of several orbiters, which then relay it to Perseverance. (Perseverance can receive direct comms from the DSN as a back up, but the data rate is slower.)

AI Enhances Mars Rover Navigation

In this demonstration, the AI model analyzed orbital images from the Mars Reconnaissance Orbiter’s HiRISE camera, as well as digital elevation models. The AI, which is based on Anthropic’s Claude AI, identified hazards like sand traps, boulder fields, bedrock, and rocky outcrops. Then it generated a path defined by a series of waypoints that avoids the hazards. From there, Perseverance’s auto-navigation system took over. It has more autonomy than its predecessors and can process images and driving plans while in motion.

There was another important step before these waypoints were transmitted to Perseverance. NASA’s Jet Propulsion Laboratory has a “twin” for Perseverance called the “Vehicle System Test Bed” (VSTB) in JPL’s Mars Yard. It’s an engineering model that the team can work with here on Earth to solve problems, or for situations like this. These engineering versions are common on Mars missions, and JPL has one for Curiosity, too.

“The fundamental elements of generative AI are showing a lot of promise in streamlining the pillars of autonomous navigation for off-planet driving: perception (seeing the rocks and ripples), localization (knowing where we are), and planning and control (deciding and executing the safest path),” said Vandi Verma, a space roboticist at JPL and a member of the Perseverance engineering team. “We are moving towards a day where generative AI and other smart tools will help our surface rovers handle kilometer-scale drives while minimizing operator workload, and flag interesting surface features for our science team by scouring huge volumes of rover images.”

AI is rapidly becoming ubiquitous in our lives, showing up in places that don’t necessarily have a strong use case for it. But this isn’t NASA hopping on the AI bandwagon. They’ve been developing automatic navigation systems for a while, out of necessity. In fact, Perseverance’s primary means of driving is its self-driving autonomous navigation system.

One thing that prevents fully-autonomous driving is the way uncertainty grows as the rover operates without human assistance. The longer the rover travels, the more uncertain it becomes about its position on the surface. The solution is to re-localize the rover on its map. Currently, humans do this. But this takes time, including a complete communication cycle between Earth and Mars. Overall, it limits how far Perseverance can go without a helping hand.

NASA/JPL is also working on a way that Perseverance can use AI to re-localize. The main roadblock is matching orbital images with the rover’s ground-level images. It seems highly likely that AI will be trained to excel at this.

It’s obvious that AI is set to play a much larger role in planetary exploration. The next Mars rover may be much different than current ones, with more advanced autonomous navigation and other AI features. There are already concepts for a swarm of flying drones released by a rover to expand its explorative reach on Mars. These swarms would be controlled by AI to work together and autonomously.

And it’s not just Mars exploration that will benefit from AI. NASA’s Dragonfly mission to Saturn’s moon Titan will make extensive use of AI. Not only for autonomous navigation as the rotorcraft flies around, but also for autonomous data curation.

“Imagine intelligent systems not only on the ground at Earth, but also in edge applications in our rovers, helicopters, drones, and other surface elements trained with the collective wisdom of our NASA engineers, scientists, and astronauts,” said Matt Wallace, manager of JPL’s Exploration Systems Office. “That is the game-changing technology we need to establish the infrastructure and systems required for a permanent human presence on the Moon and take the U.S. to Mars and beyond.”

MicroSD Express cards are still a little hard to find, considering they’re pretty new and only really started becoming popular last year once the Switch 2 came out. These upgraded versions of microSD cards are the only ones compatible with the Switch 2 for expanding its storage, os if you’re already starting to feel the crunch on your console, it’s worth picking one up. Samsung’s P9 microSD Express card is on sale right now — you can grab the 512GB version of $80, which is 33 percent off and one of the best prices we’ve seen.

The P9 boasts transfer speeds of up to 800MB/s, making moving games to the card that much faster. As for load times, in our testing we found that any microSD Express, the standard the Switch 2 requires, will offer roughly the same performance. This format is pretty new, so there aren’t a ton of cards on the market. As such, the P9 makes our list of best microSD cards for the Nintendo Switch 2.

The P9 microSD Express is also compatible with the Steam Deck or any other gaming console that accepts the format, as well as cameras and more.

Samsung

Advertisement

Get it now for 33 percent off. 

Follow @EngadgetDeals on X for the latest tech deals and buying advice.

In the realm of first-world problems, your cheap wall clock doesn’t keep time, so you have to keep setting it. The answer? Of course, you connect it to NTP and synchronize the clock with an atomic time source. If you are familiar with how these generic quartz clock movements work, you can probably guess the first step is to gut the movement, leaving only the drive motor.

The motor is somewhat like a stepper motor. The ESP8266 processor can easily control the clock hands by sending pulses to the motor. The rest is simple network access and control. If the network time is ahead, the CPU gooses the clock a little. If it is behind, the CPU stalls the clock until it catches up.

If you’ve ever done a project like this, you know there is one major problem. At some point, the processor needs to know where the hands are now. On initial setup, you can force the issue. However, if the power goes out, it won’t work well. If the power goes out at, say, 8 AM and turns back on at 9 AM, the CPU will be happy to correct the time to agree with the NTP time. The problem is that the processor has no idea that the hands started at 8 AM, so the time will be off.

To combat this problem, the design uses an EERAM chip to store the current time. In the event of a power failure, the CPU knows where its hands are and can adjust accordingly.

While you usually use these movements to keep time, once you can control them, you can do any crazy thing you like. Or, even anything as artistic as you can dream up.

When specifically tailored queries made to test Apple Intelligence using developer tools are intentionally ambiguous about race and gender, researchers have seen biases pop up.

AI Forensics, a German nonprofit, analyzed over 10,000 notification summaries created by Apple’s AI feature. The report suggests that Apple Intelligence treats White people as the “default” while applying gender stereotypes when no gender has been specified.

According to the report, Apple Intelligence has a tendency to ignore a person’s ethnicity if they are caucasian. Conversely, any messages that mentioned another ethnicity regularly saw the notification summary follow suit.

The report found that when working with identical messages, Apple’s AI model only mentioned a person’s ethnicity as being white 53% of the time. But those figures were considerably higher for other ethnicities; their ethnicity was mentioned 89% of the time when they were Asian, 86% when they were Hispanic, and 64% when they were Black.

The research claims that Apple Intelligence assumes that the person mentioned in the messages is white the majority of the time. Effectively, the model believes that white is the norm.

Another example shows Apple Intelligence assigning gender roles when none were given.

The tests used a sentence that mentioned both a doctor and a nurse, stopping short of getting into specifics. However, Apple Intelligence created associations that weren’t in the original message in 77% of the summaries tested.

Further, 67% of those instances saw Apple Intelligence assume that the doctor was a man. It then went on to make a similar assumption that the nurse was a woman.

Notably, it’s believed that the AI’s training data led to the assumptions. They closely align with U.S. workforce demographics, suggesting that the AI is simply working with the information it was trained on.

Similar biases were observed across a variety of different criteria. The report shows that eight social dimensions, including age, disability, nationality, religion, and sexual orientation, were all subject to the AI’s assumptions.

Methods and limitations

In a report detailing its work, AI Forensics explains that it used a custom application made using Apple’s developer tools to run its tests. That application hooked into Apple’s Foundation Models framework to simulate real-world messages.

That approach means that the testing closely matches what users of other third-party messaging apps might experience. However, there is still some considerable room for inaccuracy.

AI Forensics admits that its “test scenarios are synthetic constructions designed to probe specific bias dimensions, not naturalistic notifications.”. It adds that real messages may differ in the way that they are written and, as a result, interpreted by Apple Intelligence.

The outfit also notes that real-world messages may not use the same “ambiguous pronoun references” as its test messages. This, we think, is the biggest flaw in the research.

However, it’s important to note that any biases, like the ones shown in this report, can be huge at Apple’s scale. Apple Intelligence is used on hundreds of millions of devices every day.

Similar results to those highlighted in this report may well occur in considerable numbers.

More bad press for Apple’s summaries

This isn’t the first time that Apple’s AI-powered notification summaries have come under fire. In December 2024, the BBC complained that summaries of its news articles were wrong.

One example notification read “Luigi Mangione shoots himself,” referring to the man arrested for the murder of UnitedHealthcare CEO Brian Thompson. Mangione was, and is, alive and currently awaiting trial.

Apple subsequently disabled notification summaries for news apps while it worked on fixing the issue. But this report shows that notifications for communication apps, like Messages, continue to prove problematic.

Apple is clearly aware of Apple Intelligence’s shortcomings. The company recently signed a deal with Google to bring its Gemini AI model to Siri.

But following reports that the revamped Siri will not ship with iOS 26.4 as expected, hopes of an imminent improvement have been dashed.

Interestingly, AI Forensics also notes that Google’s Gemma3-1B model is much smaller than Apple’s, yet more accurate. In testing, it hallucinated

less frequently as well as less stereotypically.

Apple recently placed software chief Craig Federighi in charge of its AI efforts, a sign that it isn’t happy with Apple Intelligence as-is. But improvements are slow to come.

Hope of a quick fix for the kinds of biases highlighted by AI Forensics is likely to be dashed much more quickly.

• Harvard engineers created robotic muscles using rotational multi-material 3D printing techniques
• Hollow polyurethane tubes filled with air or fluid allow pre-programmed movement
• A spiral actuator unfurls while a gripper curls fingers around objects

A team of engineers at Harvard has developed a 3D printing technique that allows fully flexible structures to twist, bend, or lift on demand, creating what researchers describe as robotic “muscle.”

The method, called rotational multi-material 3D printing, merges several printing methods and enables the simultaneous deposition of multiple materials through a single nozzle that rotates continuously while printing.