Do robot pool cleaners clean walls and the waterline, or just the floor?

The split between Level II and Level III in pool robots is cleaner than in almost any other robot category. Level II pool robots share a defining characteristic: their navigation treats the pool as a single flat surface. The Maytronics Dolphin E10 at $549 exemplifies this architecture. Its single-motor random-path system is documented as covering pool floor without mapping or adaptive routing. Operator sets the filter run duration and deploys the unit; cycles run deterministically. The robot never climbs the wall because its drive system is not engineered for it.

The Maytronics Dolphin Escape, a $799 unit built for above-ground pools, takes a different architectural approach with its HyperGrip track design specifically engineered for vinyl walls where standard wheels slip or cause damage. Maytronics documents wall-climbing capability for the Escape. The catch, which Maytronics also documents, is that the Escape still runs random-path navigation without mapping. Wall climbing without mapping means the robot may cover some wall sections multiple times and miss others entirely, because it has no model of where it has been.

That architectural gap explains why both robots sit at Level II. Autonomy classification in this system is not about whether a robot can physically reach a surface. It is about whether the robot can execute a complete, repeatable cleaning cycle across that surface without the owner compensating for routing failures. Random-path navigation, by definition, requires the owner to accept probabilistic coverage rather than reliable coverage, which keeps the human in a supervisory role even when the robot is running unattended.

Level III pool robots share two documented capabilities that their Level II counterparts lack: systematic navigation (typically mapping-based) and multi-surface routing that covers floor and walls within a single cycle. The Aiper Scuba S1 sits at the boundary. It is classified at Level II, with manufacturer documentation noting WavePath Navigation 2.0 using inertial, acceleration, and infrared sensors for systematic floor coverage with wall-climbing capability to the waterline. Owner reports from pool forums in 2024 document wall-climb speed at roughly 10 feet per 5 minutes ascending. The autonomy classification reflects that the S1 lacks mapping: its systematic floor coverage does not persist as an internal model of pool geometry, and the operator cannot inspect or correct a coverage map. The robot runs, finishes its battery, and is retrieved. That cycle still places the owner in a scheduling role.

The Maytronics Dolphin Liberty 300 at $1,249 crosses into Level III via its documented smart-mapping capability. Maytronics documentation lists floor, wall, and waterline coverage paths enabled by a stored map of pool geometry. The Liberty 300 is cordless, using a magnetic charging dock, which eliminates the cable-management friction present in tethered Level III designs. Its autonomy ladder justification, per Robovations classification, notes that waterline and corner coverage remain dependent on algorithmic design rather than adaptive correction. That distinction matters: the robot follows a pre-computed path to the waterline, but it cannot recognize that it missed a section and revisit.

The waterline band is where Level III claims diverge most from Level III evidence. Maytronics, Beatbot, and Aiper all document waterline coverage at their flagship Level III price points. The substantive question for owners is what that documentation actually covers: whether the robot’s brushes make sustained contact with the meniscus on the pool-surface material in their specific pool, or whether the coverage claim is satisfied by the robot’s drive train reaching the geometric waterline without the brush geometry pressing flat against the pool’s curving surface at the water’s edge.

The price jump from Level II to Level III in pool robots is not primarily about motor power, filter capacity, or waterproofing depth. It is about the sensor and compute architecture required to build and execute a coverage map. The Polaris Alpha iQ+ at $1,649 uses dual sonar mapping with 360-degree obstacle detection and AI-optimized path planning, per Polaris manufacturer documentation. That sensor array is the cost driver. A Level II robot at $549 runs a drive motor, a suction motor, and a timer. A Level III robot runs all of that plus a spatial model, a sensor array to build it, and software to route the drive motor through a coverage sequence that closes on the whole pool surface.

The Beatbot AquaSense X at $4,250 represents the current high end of this architecture: camera and ultrasonic array with AI obstacle mapping and adaptive pathing, per Beatbot manufacturer disclosures and CES demonstration coverage. Battery capacity is documented by Beatbot at 360 Wh with up to 600 minutes of runtime. The autonomy classification remains Level III because field evidence on seasonal variation and obstacle-avoidance reliability across pool types is not yet established. The robot’s ceiling is documented; its performance floor across diverse pool geometries is not.

Pool geometry matters here in a way that distinguishes this category from indoor vacuums. Indoor SLAM systems work in stable environments where walls stay put. Pool geometry varies seasonally as water level changes with evaporation and rain, varies by material (gunite has different friction than fiberglass), and has no standardized shape. An irregular pool with a spa attachment, a shallow beach entry, and a gradual slope is a fundamentally different mapping problem than a rectangular 15×30 pool. Most Level III manufacturer documentation is verified against rectangular in-ground configurations, per product spec sheets.

Owner reports aggregated on pool-cleaning forums describe a consistent pattern of waterline shortfall even on robots with documented three-surface coverage. Three failure modes recur. First, robots that reach the waterline on one pool surface material (gunite or plaster) but lose traction on another (fiberglass or vinyl) before completing the waterline circuit. Second, robots that make mechanical contact with the waterline but with insufficient brush pressure to remove established calcium deposits, which require pre-treatment with a calcium remover before the robot’s brushes become effective. Third, robots that clean the waterline on initial deployment but do not re-prioritize the waterline band after algae has re-established during a multi-day interval between cleaning cycles.

None of these shortfalls disqualify a Level III classification. Level III (Conditional Autonomy) requires end-to-end task completion without teleoperation within the operating design domain. If the operating design domain is a rectangular gunite pool with moderate debris load, and waterline coverage in that domain is documented as functional, the classification holds. The gap is between the operating design domain in the manufacturer’s documentation and the pool configuration the owner actually has.

The cordless designs add a second variable. A robot that holds against a vertical wall through impeller suction or tracked grip loses margin as the filter canister fills during a cycle. Owner reports describe a waterline pass late in a run that reaches the surface less consistently than the early-cycle demonstration, because the same motor is now lifting a heavier debris load while climbing.

• Kidney-shaped and freeform pools: curved geometry challenges systematic wall routes designed for flat rectangular surfaces.
• Beach entries and gradual slopes: the floor-to-wall transition is ambiguous; robots optimized for a 90-degree angle lose traction on gradual slope approaches.
• Spa attachments: most pool robots document capability for one contiguous pool volume; attached spas require a separate deployment cycle.
• High waterline calcium in hard-water regions: brush pressure documented for routine maintenance loads does not scale to established calcium lines without chemical pre-treatment.

These are not manufacturing defects. They are constraint boundaries that sit outside the operating design domain the classification documents. An owner whose pool matches the documented domain will find the Level III coverage claims accurate. An owner whose pool does not match that domain will find the waterline claim is a marketing statement rather than a performance guarantee.

Cordless and corded designs split the Level III field along a maintenance axis that is separate from coverage. A tethered robot draws continuous power, so suction and wall adhesion do not decline across a cycle, but the cable twists and needs untangling between runs, and on a freeform pool the tether length limits how far the robot reaches before it must turn back. A cordless robot removes the cable and the snag risk, but caps a cleaning session to a battery charge, and on a larger pool it may exhaust that charge during the floor pass before it reaches a thorough waterline circuit.

Neither architecture is inherently higher on the Autonomy Ladder. The Ladder measures whether a robot completes its coverage domain without intervention, not how it is powered. A cordless Level III unit and a tethered Level III unit occupy the same rung when both close on the full pool surface. They differ in the owner effort each demands between cycles, cable handling on one side and recharge scheduling on the other, which the Robovations maintenance score records separately from the autonomy classification.