Apple’s current SVP of hardware engineering John Ternus will take over as the new CEO when Tim Cook steps down this September. Cook said in a statement: “It has been the greatest privilege of my life to be the CEO of Apple and to have been trusted to lead such an extraordinary company.”
Following the death of co-founder Steve Jobs, Cook led the charge for Apple’s post-iPhone and iPad era, launching the AirPods, Apple Watch and Vision Pro. He also turned the company into a service provider with the launch of Apple TV, Apple Music and several other subscription services. Cook will transition to a new role as executive chairman of Apple’s Board of Directors.
Ternus joined Apple in 2001 and became VP of hardware engineering in 2013, later transitioning to a senior executive role in 2021. You might have spotted Ternus being prominently featured at the MacBook Neo launch a few months ago. Expect to see a lot more of him.
We’ve seen a few beautiful moments from the Artemis II crew’s history-making trip around the Moon. Now, Reid Wiseman, the mission’s commander, has something to share. While mission specialist Christina Koch was using a Nikon camera to snap stunning still images of the Earthset, Wiseman used an iPhone 17 Pro Max to film the moment. “I could barely see the Moon through the docking hatch window, but the iPhone was the perfect size to catch the view.”
This was the first time that human eyes had witnessed an Earthset in 54 years, since the Apollo 17 mission.
This year's edition of the robot half-marathon hosted more than 100 competitors, with first place going to Honor and its red-clad robot named Lightning. Last year's event featured many bipedal robots receiving assistance from human operators who ran alongside them, along with some comical mishaps. According to the BBC, around 40 percent of the robots competed autonomously this year, while the rest were remote-controlled.
The PencilVac is light, mobile and easy to use, making it great for smaller living spaces and quick clean-ups. However, it struggles with thicker carpet and rugs. It could be perfect for a future-forward witch costume next Halloween, though.
Smart home devices have come a long way from the plain white boxes we once hid behind sofas. Voice assistants sit openly on shelves now, and small robotic helpers are slowly making their way into living spaces. For all their usefulness, though, most still feel more like appliances than companions. They respond when spoken to, perform tasks, then go quiet, making the whole relationship feel transactional rather than warm.
Samsung Design seems to think there’s a better way. At Milan Design Week 2026, its Open Lab unveiled the AI Companion, a small spherical robot designed to feel less like a gadget and more like a genuine presence. The concept frames these companions as friends that “understand you and grow with you,” bringing delight and warmth to daily life rather than simply waiting for the next voice prompt.
Designer: Samsung Design
The AI Companion’s form is its first deliberate statement. It’s a near-perfect orb, compact and smooth, with a presence that feels more like a creature than a consumer device. There are no sharp edges, glowing rings, or intake vents, none of the usual signals of smart home hardware. What it has instead is a small circular screen that reads as expressive eyes, giving it a quiet, almost attentive quality.
That face is where the design becomes truly surprising. The upper section of the sphere lifts open, almost like a creature raising its head, to reveal a compact projector tucked inside. It’s a small mechanical gesture that carries outsized meaning. The transition from sealed orb to open, projecting device doesn’t feel like pressing a button; it feels like watching something wake up and decide to share a moment with you.
With that projector now exposed, the AI Companion can cast games, animations, and interactive content directly onto the surface in front of it. The experience shifts from a one-on-one interaction to something more communal, turning a tabletop into a small shared stage. It’s the kind of feature that makes the device feel genuinely social, designed for moments between people rather than a single user quietly issuing voice commands.
Part of what makes the AI Companion feel so considered is how personality has been worked into its physical design. It comes in distinct variants, each with its own visual character, from a minimal white orb to one with a yellow cap-like shell to another wrapped in teal and rust-orange. These aren’t cosmetic afterthoughts; they suggest that each companion is meant to reflect the personality of whoever it lives with.
Samsung Design also sees these companions as inherently social. They can interact with each other, creating the kinds of playful exchanges that make them feel more like characters sharing a space than devices sitting on a shelf. The AI Companion is explicitly a concept and isn’t headed for retail, but it lays out a compelling vision for home AI that’s designed to be felt, not just heard.
To make up for an incredibly laughable inaugural event, Beijing is running back its humanoid robot half-marathon. Fortunately, the event that pits humanoid robots made by Chinese companies against each other across 13 miles went a lot smoother this year.
This year's half-marathon hosted more than 100 competitors, with first place going to Honor, better known for its smartphones, and its red-clad robot named Lightning. Living up to the name, the gold medalist finished the race in 50 minutes and 26 seconds. That's several minutes faster than the human record that was recently set by Uganda's Jacob Kiplimo last month.
Honor swept the other podium spots, with the important caveat that they all navigated the course autonomously, according to the state-sponsored television news agency CCTV. That's a massive improvement over last year, where the fastest time among 21 robots was achieved by Tiangong Ultra with a record of two hours and 40 minutes. Last year's event saw many of the bipedal robots receiving assistance from human operators who ran alongside them, as well as some comical mishaps, like falling at the starting line.
However, the BBC reported that around 40 percent of the robots competed autonomously this year, while the rest were remote-controlled. Despite the rapid improvements, this year's event still had its fair share of crashes, even from Honor's robots.
This article originally appeared on Engadget at https://www.engadget.com/ai/beijings-robot-half-marathon-is-back-for-its-second-year-with-far-less-embarassing-results-191308396.html?src=rss
For most people, saying something as simple as “good morning” to a stranger or asking for directions takes no effort at all. For the tens of millions worldwide who live with speech impairments or are completely mute, those same moments can be frustrating or simply inaccessible. The tools that exist to help, from apps to letter boards, tend to make communication slower rather than simpler.
That’s what designer Ivana Nedeljkovska set out to change with Your Voice, an assistive communication concept built on a simple premise: the body already tries to speak, even when no sound comes out. Rather than adding yet another screen or typing interface to the equation, the system works with what the body naturally does, turning the attempt to communicate into communication itself.
Your Voice consists of two components. A flexible patch worn on the neck detects the muscular movements the body makes during attempted speech, even when the vocal cords produce no sound at all. Those signals are transmitted in real time to a small, spherical robotic unit, which converts them into audible speech. The patch reads the intention; the robot gives it a voice.
What that means in practice is the removal of the pause that defines most assistive communication right now. Someone with a speech impairment attending a meeting doesn’t have to look away from the conversation to type out a response. A child who can’t speak can call for a parent without reaching for a device first. The thought and the response happen almost simultaneously.
The robotic unit’s form was guided by Nedeljkovska’s early inspiration from an orange, its rounded shape steering the design away from anything clinical. The polished sphere, embedded display panel, and mesh speaker grilles give it a refined look that doesn’t betray its purpose at a glance. It’s something you’d carry without self-consciousness, which matters more in assistive technology than it’s often given credit for.
The display panel on the robot unit adds another layer to the audio output. It shows transcribed words in real time so conversations can continue even in noisy environments or when someone nearby can’t quite hear what was said. The neck patch is designed to sit against the skin comfortably for extended wear, and the robot is compact enough to be held in hand or placed nearby.
Most assistive communication tools are designed around output: a screen to tap, an app to navigate, a board to point at. Your Voice flips that logic by making the body the input. That shift in thinking is arguably the most significant thing the concept offers, more so than any single feature, because it treats a physical limitation as a starting point rather than a constraint.
It’s still a concept, and turning neck muscle signals into reliable speech at scale is a complex engineering challenge. But the direction Nedeljkovska points toward, communication that asks nothing extra of the person trying to be heard, is one that the assistive technology field sorely needs. The ambition isn’t simply to build a better device; it’s to stop making communication feel like work.
NASA spent $2.7 billion building Perseverance and getting it to Mars. CircuitMess will sell you a buildable, functional, AI-capable replica for $309, and you get to solder every joint yourself. The kit launched in early 2025 and has sold 4,000 units across five restocks, each batch clearing out in approximately two hours. The math on that kind of sell-through rate points to something working at a level most STEM products never reach. The engineering decisions behind the kit explain why people keep showing up for the restocks instead of waiting for broader availability.
CircuitMess and GeeekClub secured NASA approval for the branding and matched the mechanical design to the real rover’s geometry, down to the rocker-bogie suspension that allows independent wheel movement across uneven terrain. The hardware includes six DC motors for propulsion, two servo motors for the arm and camera, a dual-core ESP32 microcontroller, and an AI-capable camera module with object recognition. Assembly involves soldering 300-plus components over roughly 20 hours, with all tools provided in the kit. Control options include a custom RF controller you build yourself, WiFi remote access, and autonomous navigation modes powered by the onboard AI. The firmware lives on GitHub as an open-source repository, and the rover accepts Python, C++, and Arduino IDE programming for anyone who wants to modify its behavior or add new capabilities through the modular expansion ports.
The process is guided but not restrictive, and after completing the base rover, users can reprogram it in Python or C++, experiment with CircuitBlocks, integrate additional modules, or alter its behavior entirely. That distinction matters because most STEM kits treat the build as a finish line. You follow the instructions, snap the final piece into place, drive it around for an afternoon, and then it becomes shelf decoration. The kit functions as a flexible platform rather than a one-time build. The modular architecture accepts additional sensor modules, letting builders upgrade and enhance their rover’s capabilities over time. The included fiducial marker cards give the AI camera immediate objects to recognize and track, so the computer vision feature has a real use case right out of the box. The orange foam cubes and balls visible in the kit photography serve as sample collection targets for the robotic arm, turning the rover into a functional system with tasks to perform, not a static display model.
While most STEM products bury their appeal in technical jargon and uninspired concepts, CircuitMess tapped into current trends to entice a broader audience, with strategic direction that aligned with Mars rover making real discoveries millions of miles away. Perseverance landed in Jezero Crater in February 2021, and for a sustained period afterward, space exploration felt culturally immediate in a way it hadn’t since the early shuttle era. People were watching a robot drive around Mars in near real-time, following sample collection updates, tracking the Ingenuity helicopter flights. CircuitMess launched this kit while that public interest was still live, and the timing was surgical. The kit didn’t sell because people wanted another electronics project. It sold because people wanted to understand how the thing they were watching on the news actually worked, and CircuitMess offered a credible path to that understanding.
The NASA Mars Perseverance Rover kit is available at circuitmess.com for $349. Since launching in early 2025, the kit has sold 4,000 units across five restocks, with each batch selling out in approximately 2 hours. If you’re the kind of person who followed the Perseverance mission beyond the landing headlines, who knows what a rocker-bogie suspension does or why a dual-core processor matters for onboard AI, this is one of the few educational kits that respects that level of interest.
There's just something magical about a robot that can convert into a car, tank or plane. It seems that Hollywood agrees as there are several major franchises based around that concept. As someone who grew up in the 80s and 90s, Transformers hold a special place in my heart, despite Michael Bay's best efforts at tarnishing its legacy. I spent countless hours as a kid playing with Hasbro and Takara's plastic figures, but there was one type of toy I always wanted but never got: a robot that could transform on its own just like the ones I watched on TV. That changed a few years ago when Robosenlaunched its line of officially licensed auto-converting models, and from what I've seen, its latest release featuring Soundwave might be its best yet.
Design: More than meets the eye
As a follow-up to previous bots featuring Optimus Prime, Megatron, Bumblebee, Grimlock and others, Soundwave was a superior choice, and Robosen has done a more than respectable job of bringing him to life. Not only can he spit out classic lines performed by original voice actor Frank Welker, both his robot and alt modes are a vision straight out of the first-generation (G1) cartoon. Everywhere you look, there are a ton of lovingly crafted details like the working eject button for the cassette slot and all sorts of lights. Robosen's head sculpt is spot on, and it even includes additional LEDs for his eyes and shoulder cannon. Granted, there is a bit of kibble (aka what fans call out of place parts leftover from transformation), like hands that don't properly fold away when Soundwave turns into a boombox, but that's really nitpicking. Between his incredibly accurate design, vocoder-powered vocals and an imposing stature that stands at around 14 inches tall, there's no way you can call this rendition of Soundwave uncharismatic.
A simply superior head sculpt.
Sam Rutherford for Engadget
However, the real magic happens when you turn him on (there's a little button on his back) and say "Hey, Soundwave." From here, you can use more than 50 different voice commands to boss him around like you're the leader of the Decepticons. This includes asking him to say iconic lines, respond to an Autobot attack or just wishing someone a happy birthday. Naturally, the most impressive request is asking him to transform, at which point 28 high-precision servo motors and multiple motion sensors coordinate. This allows Soundwave to convert from boombox mode to robot and back again, complete with the required sound effects. Even as a jaded adult, there's still something incredibly enchanting about watching a Transformer actually transform on its own. But that pales in comparison to the one-of-a-kind reaction my four-year-old son gave me when I repeated the process for him. There was a joy in his face I'm not sure a grown-up can truly express, as he gets to experience this without knowing this bot costs a cool $1,400.
Robosen's free mobile app features a D-pad so you can easily tell the robot where to walk.
Sam Rutherford for Engadget
While testing Soundwave's various commands, I did notice that his voice recognition can be somewhat hit or miss. I found that even a little background noise can cause issues. To be consistently heard, you have to speak louder than you think you should. The real key is being very deliberate with a sharp "Hi" or "Hey" to activate Soundwave's wake phrase properly. Alternatively, if you prefer not to yell at your robots, there's also a free companion app that allows you to send commands by simply pressing a button, which was super easy to set up and quickly became my preferred control scheme.
Here is Robosen's version of Soundwave (right) compared to the Studio Series 86 figure (left).
Sam Rutherford for Engadget
In addition to making it easier to get Soundwave to walk around (it's much more fun to use a virtual D-pad than yell "Walk forward" all the time), the app also provides a more straightforward way of discovering what he can do while reducing the ambiguity of voice commands. There are handy buttons for all his voice lines and poses, plus there’s a toolkit for creating some of your own. You can also download more from the app, though there weren't any for me to test out because Soundwave wasn't officially out yet at the time of writing. There's even a Mini Theatre mode that allows the bot to perform short skits, and if you're lucky enough to own some of Robosen's other Transformers toys, like Megatron, some of these scenes can even be performed in tandem.
I don't think any Autobots have a chance against a Soundwave this big.
Sam Rutherford for Engadget
One awkward thing about Roboen's more sophisticated approach to toy robots is that Soundwave loses some of his structural integrity when his motors are off. For example, when you power him down in robot mode, he bends over backwards and gets stuck halfway between his humanoid and boombox forms. I assume this is to prevent him from falling over, which is a good thing; it just looks kind of weird. On the flip side, if you pick him up while in stereo mode, his limbs tend to droop. However, perhaps the biggest downside to Soundwave is one inherent to his design. Because his alt mode is a boombox instead of a vehicle like Optimus, Bumblebee and others, he can't pull double duty as a remote control car. But what Soundwave lacks in mobility, he makes up for with his signature acoustic skills.
Audio: Not just a bot, he's a real boombox too
From the front, Soundwave's boombox alt mode looks damn near perfect.
Sam Rutherford for Engadget
Soundwave turning into a boombox that can't play music just wouldn't make sense. Thankfully, that's not an issue as this bot's buttons aren't just for show. Hitting Play lets you listen to original tracks from the G1 cartoon, complete with the ability to pause or skip to the next track. You can also hold the record button to save a personal message for later, though I found this feature has a bit of a learning curve as Soundwave tends to cut out one or two seconds from the beginning and end of a clip.
Inside the app, there's also a big list for all of Soundwave's voice lines and poses.
Sam Rutherford for Engadget
Most importantly, if you want Soundwave to play other tunes, you can pair it with your phone or pretty much any other mobile device and use him just like a typical Bluetooth speaker. Now it probably won't be a surprise when I say that Soundwave's audio quality is mediocre at best. With all the various sensors, motors and moving parts, there probably isn't a ton of room for fancy drivers, so things sound tinny and flat. But in a way, that's kind of endearing because the vast majority of portable speakers back in the 80s didn't sound great either. The one thing I wish Robosen had included was a proper cassette player to really capitalize on Soundwave's classic audio capabilities. That said, even though I still have stacks of CDs and DVDs in my house, I don't have any tapes (despite their resurgence), so I get why that feature didn't make it.
Battery life
It may not be period accurate, but the addition of a USB-C port around back for power is a really nice touch.
Sam Rutherford for Engadget
Soundwave comes with a built-in 1,650mAh battery which takes about 120 minutes to charge from dead to full while offering a standby time of around 60 minutes. During my testing, I found you can get a solid 20 to 30 minutes of playtime out of him, which felt like plenty. Of course,that depends a ton on how much moving around you tell him to do. And while it certainly isn't period authentic, I really appreciate the inclusion of a USB-C port for charging.
Wrap-up
The funny thing about Robosen's Soundwave is that a toy like this would have been priceless to me as a child. But now that I'm older and I have to attach a value that goes beyond its basic price, things are a lot trickier.
I love Robosen's attention to detail. The figure looks incredible and getting voice lines from the original actor shows there's more than meets the eye to the robot’s design. But most importantly, seeing Soundwave transform on his own and stomp around like he does in the show will never get old.
As you'd expect from a toy this expensive, Robosen's packaging is excellent.
Sam Rutherford for Engadget
On the other hand, $1,400 can buy the whole family a nice three-day vacation or more than two dozen regular Transformers toys. That kind of math makes it difficult to add this Cybertronian to the household register. But for anyone who has a budget similar to a Michael Bay movie, this take on Soundwave really does feel like a dream come true. Aside from some of Robosen's other products, this robot is certainly made of sterner stuff.
This article originally appeared on Engadget at https://www.engadget.com/general/robosen-soundwave-review-a-childhood-dream-made-real-120000804.html?src=rss
RAI (Robotics and AI) Institute has built a new bipedal-wheeled robot prototype that gives us a glimpse of what versatile legs with efficient wheels can be. Designed for multi-mode locomotion, the Roadrunner weighs roughly 15 kilograms (33 pounds). Owing to its design and configuration, the robot on wheels with a set of symmetric legs can switch mobility modes based on the navigation requirement.
Roboticists are already racking their brains on developing humanoids to get work done in households. Now, with the possibility of a robot like the Roadrunner, we are definitely headed in the direction of smarter robots that thrive on agility and dexterity, and are designed to go where it’s dangerous for humans to venture. A few robotic options, in the shape of animals et al., that have semantic understanding of their surroundings have been around for a few years, but the bipedal-wheeled robot is really in a different league.
The brainchild of the Massachusetts-based institute, the Roadrunner robot is in the prototype stage, but it has moves to impress. If you don’t believe my word, take a look at the video (above) doing rounds on social media. The robot is designed to seamlessly switch between side-by-side and in-line wheel modes and stepping configurations, based on the environment it is navigating. “A single control policy” is “trained to handle both side-by-side and in-line driving,” RAI informs.
From the demonstration video, you can easily figure out the remarkable versatility of the Roadrunner. The combination of the balance on its legs and the efficiency of its wheels really allows us to watch the robot’s multi-modal locomotion properties being pulled off in style. This is possible because of the robot’s symmetrical knee joints and legs that permit it to avoid obstacles easily.
Roadrunner is able to effortlessly stand up from various ground positions and walk or roll on its wheels. It can step over obstacles with the same convenience. What really blew my mind is the robot’s balance; the ease with which it can stand up on a single wheel. All of these were successfully deployed zero-shot on the hardware,” performing every task without specific prior training for them.
RAI Institute plans to use the Roadrunner as a research platform, an agile and dynamic option to legged humanoid robots that have their restrictions with pace and mobility. RAI, founded by Marc Raibert, the man behind Boston Dynamics, has built its image in agile and highly dynamic robots, and the Roadrunner is a reflection of this. If the robot in the prototype stage is apt enough to leave us all impressed, what the advanced version would be able to pull off is anyone’s guess.
Amazon has acquired Rivr, a startup focused on autonomous robotics. Rivr is based in Zurich and was valued at $110 million in a funding round from August 2024, which both Amazon and its CEO's Bezos Expeditions participated in. Financial details of the acquisition were not disclosed.
Rivr's robots have four legs and wheels that allow it to maneuver on stairs and other potentially uneven surfaces. The company just released its second generation of the robot. The purchase will likely further Amazon's capabilities for ever-faster and more efficient package deliveries.
"This acquisition reflects our commitment to a continued investment in research, which we believe has the potential to further improve safety outcomes and the overall delivery experience for delivery service partners and their delivery associates," a representative from Amazon told The Information.
Amazon has been working toward introducing automations and robotics at various stages of its shopping business. It deployed its 1 millionth robot last summer and has future goals for automating 75 percent of all its operations.
This article originally appeared on Engadget at https://www.engadget.com/big-tech/amazon-acquires-autonomous-robotics-startup-rivr-212839750.html?src=rss
A sneaker-sized robot developed at RMIT University in Australia is making a compelling case for rethinking how humanity responds to one of the ocean’s most persistent threats. The “Electronic Dolphin” is a Wi-Fi-controlled minibot built to skim oil slicks from contaminated marine surfaces without deploying any chemical dispersants, and without putting human responders anywhere near the hazard. Detailed in the journal Small, the device is compact, remote-operated, and draws on one of nature’s more underrated structural templates to do its job. It is not the first machine built to address marine oil contamination, but it may be the first to approach the problem with this particular combination of biomimicry, material science, and autonomous ambition.
The secret is in the filter. Rather than relying on PFAS-based absorbents, which are toxic, persistent in the environment, and increasingly regulated worldwide, the RMIT team engineered a composite coating from specialized carbon layers and modified barium carbonate. The resulting material mimics the microscopic spine geometry found on sea urchins, forming tiny protrusions that trap air pockets in a precise architectural arrangement. That structure makes the surface simultaneously superhydrophobic and oleophilic, a combination that causes water to roll straight off while oil latches on and gets drawn in. The chemistry here is elegant in the way good materials science often is: solving a messy physical problem through surface geometry rather than reactive chemistry.
Designers: RMIT University
The filter sits at the robot’s nose, paired with a small onboard pump that actively draws the oil slick inward. In controlled laboratory tests, the prototype processed oil at roughly two milliliters per minute, achieving over 95% purity in the recovered material. The coating also demonstrated strong corrosion resistance when exposed to saltwater, and held up across multiple reuse cycles without meaningful degradation. Those numbers matter because reusability is one of the practical bottlenecks that has historically limited oil spill response hardware. A filter that survives repeated deployment in a corrosive marine environment is a filter worth scaling.
The current battery life runs to about 15 minutes, which is honest enough for a research prototype operating at this scale. The RMIT team is candid about the limitations, and equally clear about the trajectory. Future iterations are envisioned at dolphin scale, fully autonomous, and capable of operating in a continuous loop: skim the surface, return to a base station, drain the collected oil, recharge, and head back out. That remediation model borrows from how robotic vacuum cleaners normalized autonomous domestic cleaning, and it translates surprisingly well to open-water spill response, where the geography is hostile, the timeline is open-ended, and human supervision is expensive.
Marine oil spills remain one of the more intractable environmental disasters, not because the problem is poorly understood but because the cleanup tools available have lagged behind the scale of the damage. Dispersants break oil into smaller particles that sink rather than surface, which looks like cleanup but often relocates the harm. Booms and skimmers are manual, slow, and weather-dependent. The Electronic Dolphin does not solve all of that at once, but it represents a shift in the design logic: autonomous, chemical-free, biomimetically informed, and built from the start with continuous deployment in mind. That is the kind of thinking the problem has always deserved.
