CES is a lot of things, but intimate isn’t one of them. Global Connect solved that problem by concentrating innovation into a four-hour window at a private Las Vegas villa on January 5th (which I only later found out was Mike Tyson’s house). iMpact PR built this event for media professionals who’ve seen enough “revolutionary” prototypes fail to deliver and want extended time with robotics that might actually make it to market. The format matters because robots only prove themselves when you can test, question, and watch them handle unexpected scenarios beyond their ideal conditions.

What’s showing up covers bizarre range. AI bartenders sit alongside exoskeletons designed to reduce knee pressure by 50%. Lawn robots with four-wheel steering share space with chess-playing arms and vacuum cleaners that climb stairs. Some of these machines solve concrete problems like outdoor maintenance or mobility assistance. Others exist for entertainment, companionship, or just because someone thought teaching a robot to play chess would be interesting. iMpact PR and USA Today are co-hosting, running from 1:30pm to 5:30pm with brands that have manufacturing scale, distribution plans, and enough backing to suggest they’re serious about commercial deployment. This is where you find out whether robotics companies can deliver on their promises or just deliver good marketing.

Ascentiz BodyOS: Hip and Knee Modules

Ascentiz’s BodyOS platform takes a modular approach to human augmentation, letting users swap joint actuators depending on whether they’re climbing mountains, working construction sites, or just need help with mobility. At Global Connect, the company is demonstrating two core modules: a hip actuator that delivers 36 Nm of torque through a quasi-direct-drive system, and a knee actuator using cable-drive transmission to hit 48 Nm while keeping heavy motors away from the moving limb. The hip module can assist at speeds up to 28 km/h and reduces leg effort by 35% on inclines, while the knee module claims to cut knee joint pressure by 50% and reduce energy waste by 30%. Both modules attach to a T700 aerospace carbon fiber frame that can handle temperatures from -20°C to 60°C, making this a system designed for actual outdoor conditions rather than controlled laboratory demonstrations.

The intelligence behind these modules comes from what Ascentiz calls its “Motion Cortex,” an AI trained on over 690,000 musculoskeletal gait cycles that processes data from more than 10 sensors to recognize seven different motion scenarios with 99.5% accuracy. The system responds in under 500 milliseconds when it detects a change in movement pattern, with actuator transitions happening in under 200 milliseconds. Users can strap the entire system on in less than 10 seconds using a proprietary ETIE dial mechanism, which addresses one of the persistent complaints about exoskeletons: that they take too long to put on and adjust. The control loop runs at 20 microseconds, fast enough to respond to the unpredictable shifts in terrain and body position that define real-world hiking, climbing, or industrial work. This isn’t augmentation for the sake of looking futuristic; it’s engineering focused on whether someone’s knees will hold up after a 10-mile trek with a heavy pack.

BreakReal R1: Conversational AI Bartender

Of all the things artificial intelligence could tackle, someone finally asked the right question: what if a robot understood you were having a rough day and mixed you exactly the drink you needed? BreakReal’s R1 is billing itself as the world’s first conversational AI bartender, and yes, that means you can tell it you’re stressed about work or celebrating a promotion, and it will generate a cocktail recipe based on your emotional state and taste preferences. The system uses large language models to process natural language, so you can be as vague or specific as you want. Tell it you want “something tropical but not too sweet” or just “surprise me with something spicy,” and it will craft one of its generative recipes in about 30 seconds with ±1 ml precision. It has access to over 500 classic cocktail recipes that it can customize, and it uses optical character recognition to identify whatever spirits you have sitting on your counter, so you’re not locked into proprietary bottles or special cartridges.

The genuinely clever part is how BreakReal treats drinks as shareable data. Each cocktail the R1 creates carries emotional and taste information that can be saved to the app, shared with the global BreakReal community, and downloaded by users anywhere in the world. Someone in Tokyo could create a drink reflecting their mood on a rainy afternoon, and you could try that exact recipe in New York the next day. The system handles automatic layering for visual appeal and includes automated cleaning, which addresses the reason most people avoid elaborate cocktails at home: the cleanup is annoying. This isn’t just a novelty machine pouring pre-programmed drinks; it’s attempting to turn mixology into a social network where recipes travel across time zones and language barriers. Whether that sounds delightful or slightly dystopian probably depends on how you feel about your kitchen appliances understanding your emotional state, but it’s hard to deny the ambition of turning feelings into formulas and sharing them globally.

Artly.AI Mini Barista Bot and Bartender Bot: Precision Over Personality

BreakReal wants your robot to understand your feelings, but Artly.AI built theirs to replicate a U.S. Barista Champion’s technique with mechanical precision. The company has two products at Global Connect: the Mini Barista Bot, which has already served over 1.1 million cups since 2021, and the newer Bartender Bot, which applies the same robotics foundation to cocktails instead of coffee. The Mini Barista Bot fits into a 4×4-foot footprint and handles the complete workflow with a single robotic arm, grinding, tamping, brewing, steaming, and pouring latte art without human intervention. It’s constructed with food-grade stainless steel and modular commercial-grade components, deployed in high-traffic environments like airports, T-Mobile stores, Salesforce offices, and MUJI retail spaces where labor costs and consistency matter more than personality. Customers order via a digital kiosk and watch through transparent windows as the bot executes recipes trained by Joe Yang, ensuring cafe-quality output every time. The updated version includes a larger refrigerator, improved learning capabilities, and countertop options in Maple, White Oak, and Walnut finishes that let it blend into different decors.

At NVIDIA GTC 2025, Artly unveiled an advanced robotic hand upgrade featuring up to 20 degrees of freedom, tactile sensors, and force feedback designed to replicate human barista movements. That upgrade becomes critical for the Bartender Bot, which tackles the more intricate manipulation cocktails demand: grabbing bottles, measuring pours, shaking drinks, and adding garnishes with precision. Both bots run the same proprietary AI system combining real-time motion planning, computer vision, sensor fusion, and anomaly detection, but the Bartender Bot’s algorithms are refined for mixology’s complex performance requirements. While BreakReal’s R1 focuses on emotional intelligence and generating recipes based on your mood, Artly’s approach emphasizes craft preservation and mechanical consistency. BreakReal targets home users who want conversational AI; Artly targets commercial venues that need standardized quality without depending on bartender availability or skill variation. The technology is converging from opposite directions, and Global Connect puts both companies in the same room for the first time, offering a direct comparison between robots designed to feel what you need versus robots engineered to execute what you ordered with repeatable accuracy.

Dreame Zircon 2 Ultra and Zircon 2 Pro: Pool Cleaners with Laser Mapping and Auto-Docking

Pool robots typically bounce around randomly until they’ve covered most surfaces or the battery dies, whichever comes first. Dreame’s Zircon 2 Ultra skips the wandering entirely by using PulseMap technology, which combines LDS laser radar with multi-sensor fusion to build a detailed 3D map of the pool, capturing every curve and slope for real-time adaptive navigation. The QuadLift Four-Pump System with dual propulsion jets handles 7-in-1 cleaning across floors, steps, walls, waterlines, sun shelves, pool edges, and even the water surface, addressing the vertical complexity that separates pools from flat floors. At 10,000 GPH suction power (among the highest currently available), it tackles wet leaves, sand, insects, and floating debris in one pass. When cleaning finishes, users can set a preferred docking spot in the app, and the Zircon 2 Ultra surfaces there automatically for retrieval without requiring anyone to reach into the water or dive for a submerged robot.

The Zircon 2 Pro takes automation further with an optional Auto-Dock Base Station, one of the few available systems that eliminates the final manual step entirely. Once cleaning completes, the robot returns to the dock, exits the water, and begins charging without any user intervention, solving the problem of having to physically retrieve and plug in a wet, heavy device. The companion app tracks battery status in real-time, ensuring the robot stays charged between cleaning cycles. Beyond the docking automation, the Zircon 2 Pro uses PoolSense 2.0 technology with 12 precision sensors to map pools in 2D and plan efficient cleaning paths (S-shaped, N-shaped, Star, or Cross patterns) instead of random wandering. Its 8,000 GPH suction handles leaves, sand, and twigs while auto-adjusting power to extend runtime up to 4 hours per charge. DepthLink ultrasonic connection maintains real-time app connectivity even when fully submerged, allowing users to initiate spot cleaning or adjust settings without getting hands wet or waiting for the robot to surface.

Airseekers Tron Ultra: Four-Wheel Drive Heads Outdoors

Robots that climb stairs and mix drinks are impressive until you remember most people still push a mower around their yard every weekend. Airseekers already tackled lawn automation with the original Tron, and the Tron Ultra pushes that concept into more complex terrain with four-wheel drive, four-wheel steering, and the ability to handle slopes up to 85% (40 degrees). The FlowCut 2.0 system uses a dual-disc setup that covers 60% more grass per pass, cutting mowing time by 58% while mulching clippings into ultra-fine pieces that decompose quickly and reduce the need for external fertilizer. It can clear obstacles up to 2.36 inches high and moves laterally or diagonally thanks to that four-wheel steering, which also gives it a small turning radius that minimizes lawn compaction. The AirVision 2 navigation combines vision sensors and LiDAR for a 300-degree field of view, with radar ensuring it keeps working in rain, low light, or conditions where cameras alone would fail.

The practical improvements show up in the details: swappable batteries with three hours of runtime and a 2.5-hour charge, automatic returns to the charging dock, and multi-map memory that supports multiple lawns if you’re maintaining more than one property. A new beacon system eliminates signal dead zones under trees, bushes, or around structures, which addresses one of the persistent frustrations with robotic mowers that lose connection and stop mid-job. No RTK stations or boundary wires to install means setup happens faster and with less infrastructure cluttering the yard. Visual mapping and real-time path correction keep the Tron Ultra on track with higher accuracy than previous generations. Airseekers is launching this on Kickstarter in April 2026 at around $3,000, positioning it as a premium option for people with challenging landscapes who are tired of spending weekends on lawn maintenance. The technology has moved past flat suburban yards into hills, obstacles, and the kind of complex terrain that used to require human judgment.

Hengbot Sirius: Your Personal Robot

After a parade of robots that clean pools, climb stairs, and carry heavy loads, Hengbot Sirius takes a different approach by asking what happens when a robot’s main job is just being there. This AI-powered companion is marketed as a silent presence for people living alone, a playmate that can mimic your pet’s expressions and sounds after you upload them, or a first-person perspective camera that lets you interact with your dog remotely through the robot’s “eyes.” The pitch shifts depending on who’s buying: for couples it’s a romantic surprise delivery system, for parents it’s an educational Christmas gift and “encyclopedia of robots,” and for animators it becomes a creative platform that serves as your first robotic carrier for bringing characters to life. The flexibility is the point; Sirius positions itself as whatever you need it to be in that moment.

The customization goes deeper with DIY skins that can be 3D printed, turning the robot into a unique object that belongs exclusively to you rather than looking like every other unit off the assembly line. For engineers, an EDU version provides algorithm access, API interfaces, and one-on-one technical guidance for implementing and deploying new ideas, which transforms Sirius from a consumer product into an open development platform. This isn’t solving a concrete problem like lawn maintenance or knee pain; it’s addressing the mushier territory of companionship, creativity, and experimentation. Whether that resonates depends entirely on whether you see value in a robot that exists more for emotional presence and customizable interaction than task completion. In a showcase dominated by machines designed to do specific jobs, Sirius stands out by being deliberately vague about what it’s supposed to accomplish, leaving that definition up to whoever owns it.

SenseRobot Chess: When Robots Master the Board

Chess has been testing artificial intelligence since Deep Blue beat Kasparov in 1997, but nobody put a physical robotic arm across the board from you until recently. SenseRobot claims to be the world’s first company mass-producing AI-powered robotic arms for home use, and their lineup includes multiple chess-playing models designed for different skill levels and budgets. The Chess MINI offers an entry point for casual players, while the Lite version expands to handle both chess and draughts (checkers), broadening the appeal to families who want more than one game option. Each model integrates AI vision, robotic arm precision, and intelligent decision-making systems that have been optimized to keep production costs consumer-friendly while maintaining home-level safety standards.

What makes these systems compelling isn’t just that a robot can play chess (software has beaten grandmasters for decades), but that it physically picks up pieces, moves them, and responds to your moves in real time using AI vision to track the board state. The company offers AI game review features and multiple difficulty levels, letting beginners learn strategy against a forgiving opponent or advanced players face genuinely challenging competition. SenseRobot also supports Gomoku (a strategic board game also known as Five in a Row), giving these robotic arms versatility beyond just chess. The physical interaction matters; watching a robotic arm deliberate, reach across the board, and execute its move creates an engagement that screen-based chess can’t replicate. This positions SenseRobot’s products somewhere between educational tools for teaching children strategic thinking and entertainment for adults who want their technology to exist in three-dimensional space rather than behind glass.

The post AI Bartenders, Pool Cleaners, And Chess Arms: The Sleekest (And Smartest) Robots At CES 2026 first appeared on Yanko Design.