This tracker follows Samsung patents that give robots better balance and recovery, teach appliances to sense spills, boiling water, or mineral buildup, and add fallback controls for when a smart home glitches. Together they point to a home where machines monitor themselves and adjust instead of waiting for a command.
57 filings · tracking since May 2026 · latest Oct 2026 · updates weekly
based on all tracked filings in this watchlist · refreshes every week
Samsung has filed 50 patents around home robots, covering everything from how a robot finds its way around a room to how it docks, charges, and cleans different floor types.
The filings lean heavily on two problems: getting robots to navigate and map spaces more reliably, and making docking and charging work without the robot getting stuck or connecting badly.
a dated entry each week this watchlist moves · older entries stay archived
This week's filing focuses on keeping wheeled robots stable when a wheel loses contact with the ground. Samsung is working on ways to help its home robots stay on track even when the floor is uneven.
This week's filings focus on making a home robot more self-sufficient: finding its own charger, warming water before it arrives, tidying its maps, and bending a finger with a single motor.
This week's filings cover two ideas: letting a phone train a robot vacuum, and making a group of robots move together at the pace of the slowest one. Both push toward robots that adapt to their situation rather than following fixed rules.
This week's filing is about robot vacuums that use your photos and AI to build better maps of your home. Samsung is exploring ways to make floor-cleaning robots smarter about understanding your space.
Both new filings focus on robots tracking attention: one decides which person a robot should look at, the other lets robots monitor each other. The week's additions center on how robots manage where they direct their gaze.
This week's filings focus heavily on docking: how robots find their base, connect to it, and get serviced there. Samsung also filed on how robots sense their surroundings and the floors beneath them.
This week's filing focuses on helping a robot understand where it is inside a building. It does this by matching the robot's own scans of a space to an existing floor plan using AI.
Samsung’s tracked filings by month · newest first
15 or more filings that month · 5 to 14 · under 5
the problems Samsung keeps filing on · each with its three newest filings · new filings join every week
Cleaning robots often miss spots, get stuck on obstacles, or start jobs without enough power. These filings cover checking battery before mopping, estimating cleaning time, deciding what to drive over, finding liquid spills, and checking whether a stain was actually removed.
Robots can freeze, wander into dead ends, or drift off course with no way to recover. These filings cover systems that shake wheels free, unfreeze a stuck robot, avoid dead-end paths, and keep each wheel at the right speed.
Grabbing objects is hard when a robot cannot see well or position its fingers correctly. These filings cover an arm that zooms in before grabbing, a hand where each finger moves itself into place, and an arm that retrains itself after a failed attempt.
A robot needs to know what is in a room before it can work safely. These filings cover choosing a best position to map a room, updating maps using plain-language descriptions, using a phone camera to build a 3D map, and figuring out what each object is used for.
Robots often fail when a plan goes wrong and they have no way to adjust. These filings cover breaking instructions into step-by-step plans, fixing failed plans, and figuring out what a person is looking at.
Docking stations and charging connections can fail or waste energy if not managed carefully. These filings cover a magnetic connection that prevents bad charging, a dock that picks its own sleep mode, and a station that removes its own mineral buildup.
A few phone photos help an image-making AI fix map mistakes, like a couch read as a wall.
The flat floor plan gains height and detail, so every room becomes a full 3D picture.
The dock peels off the dirty mop cloth and fits a fresh one, so the robot never needs your hands.
A menu is projected onto the floor and the robot watches your feet to see what you pick.
every tracked filing, month by month · counts are USPTO pre-grant publications, one per publication number (an application, not a granted patent)
US 2026/0336427 A1
Detecting when a wheel unit lifts during floor transitions, the robot compensates by spinning the lifted wheels faster or slower to realign the module and maintain steering control.
US 2026/0329302 A1
Builds on fallback controls by adding a search mode when robots can't find their charging pad where they left it, using 3D shape recognition to locate it elsewhere.
US 2026/0329283 A1
Remote preheating cuts idle time at the dock by letting the robot signal the need for a wash mid-cycle, so heated water is ready when it returns.
US 2026/0329286 A1
After mapping, Samsung's approach cleans up sensor noise by analyzing the robot's actual movement path, converting rough scribbles into usable floor plans the app can display without distortion.
US 2026/0326284 A1
One motor per finger cuts the bulk that's been the barrier to dexterous robot hands that fit normal workspaces and grippers.
US 2026/0320497 A1
Phone-based image recognition offloads identification tasks to a more powerful device, letting the vacuum handle navigation while delegating object classification to the mobile processor.
US 2026/0319510 A1
A lead robot collecting width and speed specs from followers before choosing a route prevents the trailing units from scraping walls or getting stuck on turns they can't navigate.
US 2026/0310206 A1
Sensor-derived maps get spatial details filled in using phone photos and generative AI, converting rough vacuum outlines into accurate floor plans that distinguish furniture from walls.
US 2026/0295854 A1
Eye-contact sequencing lets the robot track multiple people and decide who deserves attention next based on voice activity and body position, moving beyond the default behavior of locking onto whoever speaks loudest or stands nearest.
US 2026/0295839 A1
Peer monitoring among robots shifts safety oversight from fixed cameras to the team itself, reducing blind spots when one unit operates near another.
US 2026/0277216 A1
Detecting when the robot has been lifted off the ground lets it reduce speed on restart, preventing falls from countertops or tables where owners often place it to clean around obstacles.
US 2026/0276832 A1
Extending the sensor's usable range without losing precision on nearby obstacles, the dual-power LiDAR approach lets robots map both close furniture and distant walls in a single sweep.
US 2026/0274101 A1
Builds on the balance work by solving the docking problem autonomous robots hit during recharge cycles: the station's contacts extend on contact rather than requiring millimeter-perfect alignment from the robot.
US 2026/0272249 A1
A wheel with a narrowed middle section acts as a mechanical guide, centering the vacuum on the charging base without requiring additional sensors or camera calibration.
US 2026/0272244 A1
Ultrasonic sensing lets the vacuum detect floor type in real time without stopping, moving toward active surface monitoring instead of reactive bumping between carpet and hard floors.
US 2026/0272250 A1
After teaching appliances to sense spills and mineral buildup, Samsung now removes the final friction point: the manual pad swap. An automated dock that changes mop cloths eliminates the one truly annoying task left for humans.
US 2026/0268507 A1
A robot navigating indoors needs to know where it actually is, not just build its own map. This filing automates the step of matching a robot's self-generated layout against the building's official floor plan, eliminating manual correction work.
US 2026/0219679 A1
Dual-mode navigation lets the robot detect when its normal obstacle avoidance is cycling without progress, then switch to a different strategy to escape corners and tight spaces instead of spinning in place.
US 2026/0217147 A1
A magnetic alignment sensor confirms proper seating before power transfer begins, preventing the false-dock failures that waste battery life and leave robots stranded mid-task.
US 2026/0211425 A1
The balance-and-recovery work gains a practical complement: robots that maintain momentum through terrain shifts, reducing the vulnerable moments when switching wheel power could cause a stumble or loss of traction.
US 2026/0207024 A1
The robot now plans its mopping route based on actual battery state rather than guessing, preventing jobs that run out of power midway through a room.
US 2026/0208367 A1
Sensor-based height and geometry assessment lets the robot distinguish passable thresholds from actual obstacles, moving beyond the current binary choice between blind charging and retreat.
US 2026/0208363 A1
A camera that zooms in on the target before the gripper moves reduces the precision demands on the arm itself, letting Samsung's robots pick small objects from messy environments without requiring pixel-perfect positioning from the start.
US 2026/0202216 A1
Keeping robot maps current without manual remapping solves a practical friction point: users can now verbally update layouts rather than triggering full environment rescans, letting home robots navigate accurately after furniture moves.
US 2026/0200108 A1
Coordinating multiple camera movements from a single motor simplifies the mechanical complexity that prevents robots from tracking objects naturally. This reduces the parts count and control logic needed for fluid visual following.
US 2026/0202845 A1
Converting flat floor maps into room-height 3D models requires combining vacuum navigation data with phone camera views to fill in vertical space the robot can't see.
US 2026/0194988 A1
A foot-gesture interface replaces voice and touch inputs, letting the robot read user intent from floor-level movements. This adds a fallback control path when hands are full or voice recognition fails in noisy kitchens.
US 2026/0191385 A1
Samsung's filing lets the charging dock sleep while the robot is away, switching to a low-power state based on its own voltage readings.
US 2026/0191387 A1
A real-time computer vision model lets the robot predict task duration by analyzing room geometry and floor debris mid-clean, replacing the current guessing game for users planning around arrival times.
US 2026/0194911 A1
Vibrating the wheels to break free from snags confirms Samsung's focus on self-recovery: the robot diagnoses immobility through sensor feedback, then applies mechanical force rather than requesting human intervention.
US 2026/0182802 A1
A gripper arm lets the robot move obstacles rather than stop or route around them, solving the cleanup problem when debris blocks its path.
US 2026/0186502 A1
Mapping object function by room zone lets robots move beyond simple location awareness to understand which items they can actually manipulate, a shift from spatial awareness alone to task-readiness.
US 2026/0188149 A1
A mobile projector that navigates to a user's location and self-orients eliminates the setup burden when smart home devices need human interaction, extending the robot's practical reach beyond fixed positions or manual adjustment.
US 2026/0186505 A1
The guide robot adds personalized alerts to the balance-and-recovery work, checking your vision upfront so it can switch between visual warnings and audio or haptic cues when obstacles appear.
US 2026/0166747 A1
Samsung's filing turns a spoken command like make coffee into the exact sequence of movements a robot needs to carry it out.
US 2026/0169478 A1
Getting useful maps from a single camera angle fails when furniture blocks the view. Samsung's robot solves this by scanning multiple heights at once and selecting whichever level shows the clearest landmarks.
US 2026/0166758 A1
Self-positioning fingers let the hand adapt its grip width before closing, solving the gap between picking up thin objects and bulky ones without requiring the arm to reorient the entire hand.
US 2026/0169502 A1
A home robot that can actually operate your screens means it could handle tasks like turning off a forgotten TV without needing voice commands or app controls that already exist elsewhere in the house.
US 2026/0165599 A1
Keeping users in correct movement patterns during exercise requires active correction, not passive support, a robotic leg brace that applies targeted force through a cycling routine to retrain joint mechanics after injury.
US 2026/0166725 A1
A robot arm that learns from observation and retrains itself after failed attempts moves Samsung's home robots past one-shot programming toward machines that improve through their own mistakes.
US 2026/0166732 A1
A robot that can't find something in its mental map won't just get stuck, it'll search the home and update its understanding before trying again.
US 2026/0158650 A1
The home robots need to keep working unsupervised. This filing shows how Samsung plans to let them escape physical jams on their own, eliminating the human intervention that would otherwise interrupt a cleaning cycle or task sequence.
US 2026/0161179 A1
A robot that can hold its cleaning path around corners without overshooting or cutting inside means the machine stays in its assigned zones and covers ground efficiently rather than redoing areas or missing spots.
US 2026/0153883 A1
Cutting motor power partway through braking instead of holding full force to a stop lets robots decelerate smoothly, reducing the jerky halts that could topple cargo or spill liquids while moving through homes.
US 2026/0151914 A1
A home robot that can't see past an obstacle becomes useless. This filing shows Samsung working on a robot that physically repositions its own sensors to maintain vision around clutter, letting it keep working through the small chaos of real homes.
US 2026/0154969 A1
Samsung's home robots need to verify cleaning worked, not just assume it. This filing adds computer vision that checks whether stains actually vanished after the initial pass.
US 2026/0154645 A1
Knowing worker location in real time lets robots reroute around congestion instead of waiting for supervisors to manually reprogram paths.
US 2026/0151916 A1
A robot that tracks where you're looking lets users point at objects through eye gaze rather than speech or gestures, simplifying how people command it to fetch or interact with specific items in a room.
US 2026/0140579 A1
When smart home controllers fail, recovering control without a physical remote requires an alternative input method. Samsung's gesture-recognition backup lets users stay in command through camera-based hand signals alone.
US 2026/0131609 A1
Pivoting arm linkages let each wheel climb independently while keeping the chassis level, so a home robot maintains contact with floors and rugs without tilting or losing traction during transitions.
US 2026/0132565 A1
Keeping delicate fabrics safe while steaming wrinkles out requires precise heat control. This filing shows how Samsung splits airflow into heated and room-temperature paths so the cabinet can warm clothes selectively without risking damage.
US 2026/0131455 A1
A home robot that independently identifies and responds to liquid hazards removes one obstacle to fully autonomous cleaning, since spills currently force human intervention or risk equipment damage.
US 2026/0136438 A1
An accelerometer embedded in the cooktop glass detects boiling vibrations and cuts heat automatically, moving the timeline from sensing what's on the burner to sensing what's happening inside the pot.
US 2026/0130558 A1
A self-cleaning dock means heated-water mops stay reliable without user intervention, moving the maintenance burden away from the person who owns the robot.
US 2026/0132982 A1
Samsung's freezer-drawer filing shows the robot home needs smarter sensing before it can anticipate needs: lighting that responds to which drawer opens and time of day, not just presence. The system skips one-size-fits-all automation in favor of context.
US 2026/0132927 A1
Samsung's self-maintaining appliances need to survive dirty power. This patent adds voltage monitoring to gas burners, protecting ignition circuits from electrical spikes that shorten component life.
US 2026/0132561 A1
Getting water consumption down requires knowing individual habits first. This filing shows Samsung embedding memory into washers so machines can build usage profiles and calibrate cycles to actual household patterns rather than preset defaults.
These are patent filings, not product announcements, so they show what Samsung's engineers are exploring rather than what's shipping soon. The filings describe working solutions to real problems, like a robot finding spills or bracing itself on uneven floors, which points to active research, but a patent alone doesn't guarantee a product will reach shelves.
The filings cluster around a few recurring problems: staying balanced and mobile on real floors, recovering when controls stop responding, protecting appliances from electrical or mineral damage, and getting appliances to sense conditions like boiling water or a full freezer drawer instead of relying on fixed settings.
Yes. Alongside mobile robots like spill-cleaning bots and self-balancing designs, the watchlist includes household appliances such as washing machines, induction cooktops, gas burners, freezers, and clothing care cabinets. The common link is machines that sense a condition and adjust their own behavior rather than waiting for a person to notice.
This tracker adds new filings as Samsung publishes them, so the list grows over time rather than sitting fixed at one snapshot. We don't attach dates or promise a release schedule; check back for the latest additions to see how the pattern of filings develops.
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →