What We Learned Deploying Cleaning Robots Across 4 Las Vegas Mega-Resorts

In early 2024, I led the deployment of eight autonomous cleaning robots across four of the largest casino-resort properties in Las Vegas. Five vacuum units and three scrubber units, covering casino floors, convention centers, hotel corridors, arena spaces, and back-of-house areas.

The promise was straightforward: reduce labor costs, improve cleaning consistency, and modernize EVS operations across properties that never close. The reality was more complicated. This article documents what we learned — the good, the bad, and the things no one tells you before you sign the purchase order.

The Deployment

Installation began in February 2024 across two flagship properties, with two additional properties coming online in March. Each property presented a different challenge: massive casino floors with slot machine aisles, sprawling convention centers with reconfigurable partitions, high-traffic hotel corridors, and arena spaces that transformed between events.

EVS staff across all three shifts were trained. Adoption started slowly — as it always does with new technology — but improved as operators developed familiarity with the workflows and the robots' capabilities.

Challenge #1: The Robots Couldn't Finish the Job

The vacuum units ran approximately 3–4 hours per charge. The scrubbers lasted about 3 hours. In a facility that operates 24/7 with specific cleaning windows during graveyard shifts, this was a serious problem. Neither platform could complete large-area tasks — ballrooms, casino floors, or long corridor runs — without stopping to recharge.

This isn't just an inconvenience. When a robot stops mid-task, someone has to notice, retrieve it, dock it, wait for it to charge, and restart the job. That's labor you were supposed to be saving.

Challenge #2: Map Size Limitations Hit Hard

The mapping system calculated coverage based on bounding box dimensions (length × width) rather than actual cleanable floor space. This meant irregular layouts — T-shaped hallways, L-shaped corridors, oddly shaped convention areas — consumed far more of the system's map capacity than the real floor area warranted.

For example, one conference area measured 856' × 360' on the map (approximately 308,000 sq. ft. of bounding box), but the actual cleanable area was only 33,115 sq. ft. The system was approaching its limit on a space that was barely 11% utilized. In a mega-resort with hundreds of thousands of square feet of cleanable space, this was a fundamental constraint.

Challenge #3: "Autonomous" Still Required Constant Attention

Despite having automated docking stations for charging and draining, both robot platforms required manual intervention every 2–3 hours. Garbage bins filled up. Vacuum hoses clogged. Squeegee blades detached. Suction lines and greywater filters needed clearing.

• Vacuum units: bin capacity exhausted mid-task, hose clogs requiring manual clearing
• Scrubber units: squeegee blade detachment, suction hose clogs, greywater drain filter blockages
• One disk brush motor controller failed within two weeks of deployment
• One docking station arrived with a faulty power supply out of the box

The lesson: "autonomous" doesn't mean "unattended." Every facility manager needs to budget operator time for monitoring, maintenance, and intervention — the question is how much.

Challenge #4: Navigation Failures in Dynamic Spaces

The robots used 2D LiDAR with approximately a 10-meter (33 ft) range. In large open spaces — convention halls, wide corridors, arena floors — this was inadequate. The sensor couldn't "see" enough fixed reference points to maintain its position.

When crowds of people walked between the robot and the walls, they blocked the LiDAR entirely, causing the robot to lose localization and stop mid-task. In convention spaces where partitions moved between events, the robot's stored map no longer matched reality. QR code-based map switching, intended to solve this, frequently failed during the first month — sometimes placing the robot outside the map boundaries entirely.

Challenge #5: Zero Task Portability Between Robots

Here's one that surprised us. Cleaning tasks created on one robot could not be transferred to another robot — even when both robots shared the same map. If a robot went down for maintenance and you swapped in a replacement, every cleaning task had to be recreated from scratch.

In a single-robot deployment, this is annoying. In an eight-robot fleet across four properties, it's a serious operational risk. The cleaning path algorithm also defaulted to an outside-in pattern, which meant large areas had to be manually divided into dozens of smaller zones. One arena space required over 40 separate zones to cover properly.

What We Learned: Five Rules for Large-Scale Robot Deployment

1. Map your actual cleanable area before you buy. Bring the vendor on-site and test mapping in your real environment — not a demo room. Ask how the system handles irregular layouts and what the practical map size limits are.
2. Plan for recharge cycles. Calculate how many charging cycles a robot needs to complete your largest cleaning zone during your shortest cleaning window. If the math doesn't work with one robot, you need multiples — budget accordingly.
3. Budget operator time honestly. "Autonomous" reduces labor — it doesn't eliminate it. Plan for 15–20 minutes of operator attention per robot per shift for bin emptying, hose clearing, solution refilling, and general monitoring.
4. Test navigation in your actual operating conditions. Run the robot during peak foot traffic, not in an empty building. Move partitions, rearrange furniture, simulate real conditions. If the robot loses localization, you'll know before it's too late.
5. Evaluate fleet management capabilities. Can tasks be shared between robots? Can you manage multiple robots from a single dashboard? Can you swap a robot without rebuilding every route? If the answer is no, scale becomes painful fast.

The Path Forward: Right Robot, Right Environment

The robots we deployed weren't bad machines. In medium-sized, controlled environments they performed adequately. But in mega-resort casino properties with 24/7 operations, dynamic crowds, and massive irregular floor plans, they hit their limits hard.

The lesson isn't "robots don't work." It's "no single robot works everywhere." A hybrid fleet approach — matching robot capabilities to specific areas and tasks — delivers dramatically better results than forcing one platform to do everything.

Next-generation platforms with extended runtime, larger tank capacity, 3D LiDAR navigation, and true fleet management are now available. Facilities that struggled with first-generation robots are seeing transformational results with the right hardware matched to the right application.

Considering Cleaning Robots for Your Facility?

We've deployed cleaning robots in some of the most demanding environments in the country. We know what works, what doesn't, and how to build a program that actually delivers on the ROI promise. If you're evaluating autonomous cleaning for your facility, we'd rather you learn from our experience than repeat it.