Doosan Bobcat has unveiled a new generation of autonomous and AI-enabled construction technologies at CES 2026, headlined by the RX3 autonomous concept loader and a growing ecosystem of intelligent jobsite systems. The announcements reflect the company’s push to integrate autonomy, electrification, and artificial intelligence into compact construction equipment designed for real-world deployment.

Presented during CES Media Day in Las Vegas, the technologies are part of what Bobcat describes as a “Smart Construction Jobsite,” where machines assist operators, reduce complexity, and improve safety and productivity. While several systems remain in concept or prototype form, the company emphasized that many are moving steadily toward commercialization.

RX3 Autonomous Concept Loader

The Bobcat RogueX3, or RX3, represents the third generation of Bobcat’s autonomous loader concept. The electric-powered machine is designed to match the size and footprint of existing manned Bobcat equipment, allowing it to operate in current jobsites without major workflow changes. It uses tracked mobility to provide traction across uneven or challenging surfaces while operating quietly and without emissions.

A key feature of the RX3 is its modular design. The platform can be configured with or without a cab, equipped with wheels or tracks, and paired with different lift arms depending on the task. Bobcat said the concept could ultimately support multiple powertrains, including electric, diesel, hybrid, and hydrogen, offering flexibility as energy infrastructure evolves.

“For nearly 70 years, Bobcat has led the compact equipment industry by solving real problems for real people,” said Scott Park, vice chairman and CEO of Doosan Bobcat. “As jobsites become more complex, we’re responding with intelligent systems that help people accomplish more, faster, and smarter.”

Bobcat is also working with Agtonomy as a technology partner, using its perception and fleet management software to enable autonomous and semi-autonomous operation in agricultural and construction contexts.

AI Comes Into the Cab

Alongside the RX3, Doosan Bobcat introduced the Bobcat Jobsite Companion, described as the compact equipment industry’s first AI voice control system. Powered by a proprietary large language model running entirely onboard, the system allows operators to manage more than 50 machine functions using natural voice commands.

Operators can adjust attachment settings, engine speed, lighting, and other controls without taking their hands off the controls. Because the system does not rely on cloud connectivity, it can respond in real time even in remote or connectivity-limited jobsites.

“Jobsite Companion lowers the barrier to entry for new operators while helping experienced professionals work faster and more precisely,” said Joel Honeyman, vice president of global innovation at Doosan Bobcat.

Bobcat also announced Service.AI, an AI-powered support platform designed for dealers and technicians. The system provides instant access to diagnostics, repair manuals, service histories, and troubleshooting guidance, aiming to reduce downtime and speed up maintenance.

Safety, Displays, and Energy Systems

Doosan Bobcat showcased several additional technologies that support its smart jobsite vision. A radar-based collision warning and avoidance system uses imaging radar to monitor surroundings and can automatically slow or stop a machine to prevent accidents.

The company also revealed an advanced display concept using transparent MicroLED screens integrated into cab windows. These displays overlay 360-degree camera views, machine performance data, alerts, and asset tracking directly into the operator’s field of vision.

Powering these systems is the Bobcat Standard Unit Pack, or BSUP, a modular and rugged battery system designed for harsh construction environments. The fast-charging packs are scalable across Bobcat’s equipment lineup and are intended to support broader electrification efforts, including potential use by other manufacturers.

Toward a Smarter Jobsite

Doosan Bobcat said the technologies unveiled at CES 2026 form an integrated ecosystem rather than isolated features. By combining AI, autonomy, electrification, and connectivity, the company aims to redefine how compact equipment is operated and supported.

“We’ll combine AI, autonomy, electrification, and connectivity to create new jobsite standards,” Park said during the Media Day presentation.

While the RX3 and several systems remain concept-stage, Bobcat’s messaging at CES emphasized near-term impact rather than distant vision. The company framed these developments as practical steps toward safer, more productive jobsites where intelligent machines actively support human workers.

Mobileye has agreed to acquire Mentee Robotics in a $900 million transaction that marks a major strategic shift beyond autonomous driving and into humanoid robotics and Physical AI. Announced during CES 2026 in Las Vegas, the deal positions Mobileye to apply its autonomy technology to machines designed to work directly alongside humans in physical environments.

The acquisition combines Mobileye’s large-scale perception, planning, and safety systems with Mentee’s vertically integrated humanoid robot platform. Together, the companies aim to build general-purpose robots capable of understanding context, inferring intent, and executing tasks safely and autonomously in real-world settings such as factories, warehouses, and industrial facilities.

From Vehicle Autonomy to Embodied Intelligence

Mobileye’s core business has been built around vision-based autonomy for vehicles, with systems designed to interpret complex scenes, predict behavior, and make safety-critical decisions. Those same challenges increasingly define humanoid robotics, where machines must navigate spaces built for people while interacting with objects, equipment, and coworkers.

The company said the acquisition represents a decisive move toward Physical AI, a class of systems that not only perceive the world but also act within it reliably and at scale. Mobileye’s autonomy stack has evolved beyond navigation toward context-aware and intent-aware reasoning, providing a foundation for robots that can operate productively without constant supervision.

The move also reflects Mobileye’s effort to diversify as competition intensifies in autonomous driving and commercialization timelines extend. By expanding into humanoid robotics, the company gains exposure to a parallel growing market where autonomy software may become the primary differentiator.

Mentee’s Humanoid Platform and Learning Approach

Founded four years ago, Mentee Robotics has developed a third-generation humanoid robot designed for scalable deployment rather than laboratory demonstrations. The platform is vertically integrated, with in-house development of hardware, embedded systems, and AI software.

Mentee’s approach emphasizes rapid learning and adaptability. Its robots are trained primarily in simulation, reducing reliance on large-scale real-world data collection and minimizing the gap between simulated and physical performance. The system is designed to acquire new skills through limited human demonstrations and intent cues, rather than continuous teleoperation.

This learning framework enables autonomous, end-to-end task execution, including locomotion, navigation, and safe manipulation of rigid objects. In demonstrations, Mentee robots have shown the ability to perform multi-step material handling tasks with stability and accuracy, supporting the company’s focus on real-world utility.

Deal Structure and Commercial Roadmap

Under the terms of the agreement, Mobileye will pay $900 million for Mentee Robotics, consisting of approximately $612 million in cash and up to 26.2 million shares of Mobileye Class A stock, subject to adjustments. The transaction is expected to close in the first quarter of 2026, pending customary approvals.

Mentee will operate as an independent unit within Mobileye, allowing continuity while gaining access to Mobileye’s AI infrastructure and production expertise. First customer proof-of-concept deployments are planned for 2026, with autonomous operation as a core requirement. Series production and broader commercialization are targeted for 2028.

Mobileye said the acquisition will modestly increase operating expenses in 2026 but aligns with its long-term growth strategy.

CES 2026 and the Rise of Physical AI

Physical AI emerged as a central theme at CES 2026, with humanoid robots, service robots, and embodied AI systems moving beyond concept stages. The Mobileye-Mentee announcement underscored how autonomy is becoming a shared foundation across vehicles and robots, rather than a domain-specific technology.

Mobileye highlighted strong momentum in its core automotive business, citing an $24.5 billion revenue pipeline over the next eight years. Company executives framed the acquisition as a way to extend that success into a second transformative market without abandoning its safety-first philosophy.

“Today marks a new chapter for robotics and automotive AI,” said Mobileye President and CEO Amnon Shashua. “By combining Mentee’s breakthroughs in humanoid robotics with Mobileye’s expertise in autonomy and productization, we have an opportunity to lead Physical AI at a global scale.”

Mentee CEO Lior Wolf said the partnership accelerates the company’s mission to deliver safe, cost-effective humanoid robots capable of meaningful work in human environments.

As CES 2026 made clear, the race to define Physical AI is accelerating. With this acquisition, Mobileye signals that the next phase of autonomy may unfold not just on roads, but across factories, warehouses, and workplaces worldwide.

Samsung Electronics opened CES 2026 with a broad statement about the future of consumer technology, positioning artificial intelligence not as a feature but as the foundation of everyday living. At its annual First Look event in Las Vegas, the company introduced its “Companion to AI Living” vision, outlining how AI will connect devices, services, and experiences across the home.

Rather than focusing on a single product category, Samsung framed AI as a unifying layer across its ecosystem, spanning displays, home appliances, mobile devices, wearables, and services. Company executives emphasized that scale, connectivity, and on-device intelligence allow Samsung to move beyond basic automation toward more contextual and personalized experiences.

AI as the Core of the Entertainment Experience

Samsung’s display business showcased how AI is reshaping entertainment into a more interactive and lifestyle-oriented experience. The centerpiece of the lineup was a 130-inch Micro RGB display, which Samsung described as a major leap in screen size and color accuracy, driven by independent red, green, and blue light sources.

Supporting this hardware is Vision AI Companion, an AI system designed to act as an entertainment assistant rather than a passive interface. The system can recommend content, adjust sound and picture settings, and respond to natural language requests across Samsung’s 2026 TV lineup. AI-driven modes tailor experiences for sports, movies, and gaming, allowing users to fine-tune crowd noise, commentary, or background audio through voice commands.

Samsung also highlighted how Vision AI Companion extends beyond viewing. Users can ask for recipes based on food shown on screen, receive music recommendations to match their mood, or send content and instructions to other connected devices throughout the home. The goal, Samsung said, is to turn displays into active participants in daily routines.

Smart Homes That Anticipate Daily Needs

In the home appliance segment, Samsung presented AI-powered devices as companions that reduce friction in everyday tasks. Executives noted that SmartThings now serves more than 430 million users, giving Samsung a large data foundation to personalize experiences across households.

The Family Hub refrigerator remains central to this strategy. With an upgraded AI Vision system built on Google Gemini, the refrigerator can more accurately recognize and track food items, support meal planning, and automate grocery-related decisions. Features such as recipe recommendations, video-to-recipe conversion, and weekly food reports are designed to simplify decision-making rather than add complexity.

Samsung also showcased updates across laundry and home care. The Bespoke AI Laundry Combo removes the need to transfer loads between machines, while the latest AirDresser model uses air and steam to reduce wrinkles automatically. In floor care, the Bespoke AI Jet Bot Steam Ultra combines vision, 3D sensing, and conversational voice control to clean, monitor pets, and detect unusual activity while homeowners are away.

From Reactive Care to Proactive Wellbeing

Samsung’s long-term vision extends into digital health, where AI shifts care from reactive responses to proactive prevention. By connecting phones, wearables, appliances, and home devices, Samsung aims to detect early signs of health issues and provide personalized guidance for sleep, exercise, and nutrition.

The company described future scenarios in which connected devices suggest meals aligned with health goals, flag unusual patterns in mobility or sleep, and enable secure sharing of health data with providers through integrated platforms. Samsung also highlighted ongoing research into dementia detection, using wearables to identify subtle changes in movement, speech, and engagement over time.

Security remains a key pillar of this ecosystem. Samsung emphasized that Knox and Knox Matrix underpin its AI strategy, protecting user data across devices and continuously adapting to emerging AI-related risks.

By presenting AI as a companion woven into daily life rather than a collection of isolated tools, Samsung used CES 2026 to signal a shift toward more holistic, software-driven experiences. The company’s message was clear: the next phase of consumer technology will be defined not by individual devices, but by how intelligently they work together.

LG Electronics has unveiled its most advanced home robotics concept to date with the public debut of LG CLOiD, an AI-powered household robot designed to take over everyday domestic chores. Presented at CES 2026, the robot reflects LG’s long-term Zero Labor Home strategy, which aims to reduce the physical and mental effort required to manage a modern household through intelligent automation.

Unlike earlier home robots focused on narrow tasks, CLOiD is positioned as a general-purpose domestic assistant. LG demonstrated the robot performing a range of coordinated activities, including preparing simple meals, handling laundry from start to finish, and managing dishwashing tasks. The company says CLOiD is designed to operate as part of a fully connected home rather than as a standalone device.

Demonstrating End-to-End Household Automation

During live demonstrations, CLOiD retrieved food items from a refrigerator, placed pastries into an oven, and initiated cooking processes without human intervention. After occupants left the home, the robot was shown starting laundry cycles, transferring clothes to a dryer, and folding and stacking garments once complete. CLOiD also demonstrated the ability to unload a dishwasher and organize clean dishes.

These scenarios were designed to show how the robot understands sequences of tasks rather than executing isolated commands. CLOiD uses contextual awareness to determine when chores should begin and how appliances should be operated, adapting its actions to household routines and user preferences.

LG emphasized that the robot’s value lies in orchestration. Rather than replacing individual appliances, CLOiD coordinates them, acting as a mobile control layer that connects cooking, cleaning, and laundry into a single automated workflow.

Hardware Built for Domestic Environments

CLOiD features a wheeled base for stability and safe operation in homes with children or pets. The robot’s torso can raise or lower to adjust its working height, enabling it to reach objects on countertops, inside appliances, or closer to the floor. Two articulated arms with seven degrees of freedom each provide human-like mobility.

Each hand includes five independently controlled fingers, allowing CLOiD to grasp delicate items such as glassware as well as heavier objects like laundry baskets. LG selected a wheeled design over a bipedal form to reduce cost, improve reliability, and lower the risk of tipping during operation.

The navigation system builds on LG’s experience with robotic vacuum cleaners and autonomous home platforms. CLOiD can move smoothly between rooms, avoid obstacles, and precisely position itself for manipulation tasks in kitchens and laundry areas.

Physical AI and Smart Home Integration

At the core of CLOiD is LG’s Physical AI framework, which combines vision-based perception, language understanding, and action planning. The robot uses visual data from onboard cameras to recognize appliances, objects, and environments. This information is translated into structured understanding and then into physical actions, such as opening doors, transferring items, or adjusting appliance settings.

CLOiD’s head functions as a mobile AI home hub, housing its processor, sensors, display, speakers, and voice-based generative AI. The robot communicates with users through spoken dialogue and expressive visual cues while continuously learning household layouts and routines.

Deep integration with LG’s ThinQ and ThinQ ON platforms allows CLOiD to control and coordinate smart appliances across the home. This connectivity enables more complex automation scenarios, such as preparing meals based on available ingredients or scheduling chores around user absences.

Robotics Components and Long-Term Strategy

Alongside CLOiD, LG introduced AXIUM, a new family of robotic actuators designed for service robots and physical AI systems. Actuators control motion and force within robotic joints and are considered one of the most critical and cost-intensive components in advanced robots.

LG says its background in appliance component manufacturing provides an advantage in producing lightweight, compact, and high-torque actuators suitable for home robotics. Modular actuator designs also allow customization across different robot configurations and use cases.

Looking ahead, LG plans to expand robotics capabilities across both standalone home robots and robotized appliances. The company envisions refrigerators that open automatically as users approach and appliances that actively coordinate with home robots to complete tasks autonomously.

“The LG CLOiD home robot is designed to naturally engage with and understand the humans it serves, providing an optimized level of household help,” said Steve Baek, president of the LG Home Appliance Solution Company. “We will continue our efforts to achieve our Zero Labor Home vision.”

At CES 2026, LG positioned CLOiD as a glimpse into a future where household labor is largely delegated to intelligent machines, allowing people to spend more time on activities beyond routine chores.

Researchers at the University of Pennsylvania and the University of Michigan have developed what they describe as the world’s smallest fully programmable autonomous robots, pushing robotics into a microscopic frontier. Each robot measures roughly 200 by 300 by 50 micrometers, smaller than a grain of salt, yet integrates computing, sensing, and propulsion into a single untethered system. The robots are designed to operate independently without external control, marking a significant step forward in microscale robotics.

Unlike earlier microrobots that relied on magnetic fields or external power sources, these robots are fully autonomous. They are powered by light, which activates onboard electronics and enables them to sense their surroundings and make basic decisions. In laboratory demonstrations, the robots were able to swim in liquid environments and adjust their motion without human intervention.

The robots can be produced using established semiconductor fabrication techniques, allowing them to be manufactured at scale. Researchers estimate the cost at roughly one cent per robot when produced in large quantities. Once activated, the devices can continue operating for months, making them suitable for long-duration experiments or deployments at microscopic scales.

Autonomous Microscale Motion and Control

Movement at microscopic scales presents unique challenges because fluid resistance dominates over inertia. To address this, the robots use an electrochemical propulsion method rather than mechanical parts. By generating electric fields, the robots interact with ions in the surrounding liquid, creating movement without the need for motors or moving limbs.

This approach allows the robots to swim at speeds of roughly one body length per second. The lack of moving components makes the robots mechanically robust and resistant to damage during handling. Researchers demonstrated that the devices could be transferred between samples using standard laboratory tools without losing functionality.

The propulsion method also enables precise directional control. By adjusting electrical signals, the robots can change direction, stop, or follow preprogrammed movement patterns. This capability is essential for future applications that require coordinated motion or navigation through confined environments.

Tiny Brains and Sensing Capabilities

A key breakthrough lies in the integration of a complete computing system at such a small scale. The robots include a processor, memory, and sensors embedded directly on the chip. Power is supplied by microscopic solar cells that generate approximately 75 nanowatts under LED illumination, an extremely small energy budget compared to consumer electronics.

Despite these constraints, the robots are capable of basic sensing and decision-making. They can detect temperature changes with high sensitivity and alter their behavior in response. Researchers also demonstrated simple communication by encoding information through movement patterns that can be observed under a microscope.

These capabilities allow the robots to respond dynamically rather than follow fixed paths. While the onboard intelligence is limited compared to larger robotic systems, it represents a major step toward autonomous behavior at microscopic dimensions.

Potential Applications and Next Steps

The researchers see strong potential for applications in biomedicine, where microscopic robots could one day monitor cellular environments or deliver targeted therapies. Their small size allows them to operate in spaces inaccessible to conventional devices, including narrow fluid channels and delicate biological systems.

In manufacturing and materials science, the robots could assist in assembling or inspecting microscale components. Because the platform is compatible with standard chip manufacturing processes, it could be adapted for large-scale production and customized for specific industrial tasks.

The current demonstrations were conducted in controlled laboratory conditions, and the researchers emphasize that further work is needed to expand functionality. Future efforts will focus on improving sensing, increasing computational complexity, and enabling operation in more complex environments. Even at this early stage, the work establishes a foundation for autonomous robotics at scales comparable to biological microorganisms.

Engineers at the University of Utah have developed an artificial intelligence system that significantly improves the dexterity and intuitiveness of robotic prosthetic hands. By combining advanced sensors with machine learning, the researchers enabled a prosthesis to grasp objects in a way that more closely resembles natural human movement. The approach reduces the mental effort required by users while increasing grip precision and reliability.

For many prosthesis users, even simple tasks such as holding a cup or picking up a small object require deliberate finger-by-finger control. This added cognitive burden is one of the main reasons advanced prosthetic devices are often abandoned. The Utah team focused on restoring the subconscious, automatic aspects of grasping that most people take for granted.

Sensors and AI Enable Autonomous Grasping

The researchers modified a commercially available prosthetic hand by equipping it with custom fingertips capable of sensing both pressure and proximity. Optical proximity sensors allow the fingers to detect objects before physical contact, while pressure sensors provide feedback once an object is grasped. Together, these inputs give the prosthesis a form of artificial touch.

An artificial neural network was trained on grasping postures using proximity data from each finger. This allows the prosthetic hand to autonomously position its fingers at the correct distance to form a stable grip. Because each finger operates with its own sensor, the system adjusts all digits in parallel, producing precise and adaptable grasping behavior across objects of different shapes and sizes.

In testing, participants using the AI-assisted prosthesis demonstrated greater grip security and precision compared to conventional control methods. They were also able to complete tasks using different grip styles without extensive training, suggesting the system adapts naturally to user intent.

Sharing Control Between Human and Machine

A central design challenge was ensuring that artificial intelligence supported the user rather than competing for control. To solve this, the researchers implemented a bioinspired framework that shares control between the human and the AI system. The prosthesis assists with fine motor adjustments while allowing the user to initiate, modify, or stop actions freely.

“What we don’t want is the user fighting the machine for control,” said Marshall Trout, a postdoctoral researcher involved in the work. “Here, the machine improved the precision of the user while also making the tasks easier.”

The system blends rapid reactive responses, such as preventing excessive grip force, with higher-level planning that anticipates how objects should be grasped. This mirrors how humans naturally coordinate instinctive reactions with learned motor patterns.

Study Leadership and Future Directions

The study was led by engineering professor Jacob A. George together with Trout at the Utah NeuroRobotics Lab and was published in the journal Nature Communications. The research involved experiments with four transradial amputees, whose amputations occurred between the elbow and wrist.

Participants completed standardized dexterity tests as well as everyday activities requiring fine motor control. Tasks such as lifting a lightweight plastic cup, which require careful force modulation, became more reliable with AI assistance.

“As lifelike as bionic arms are becoming, controlling them is still not easy or intuitive,” Trout said. “Nearly half of all users will abandon their prosthesis, often citing poor controls and cognitive burden.”

George emphasized that the long-term goal is to embed intelligence directly into prosthetic devices so users can interact with objects more naturally. The team is now exploring how this AI-driven grasping approach could be combined with implanted neural interfaces, enabling thought-based control and the return of tactile sensations. By merging sensing, intelligence, and neural input, the researchers aim to make robotic prostheses feel less like tools and more like natural extensions of the human body.