Definitely Doug 2/4/22: The “Cobot” Revolution

by Doug Rice

Few challenges to today’s hospitality operations are bigger than labor. Shortages have forced many hoteliers to rethink long-held beliefs about hospitality as a business where service is provided only by smiling, welcoming humans. Hoteliers are realizing that they cannot deliver a good experience without changing the mix of humans and technology. And especially during the pandemic, we have learned that many guests want less human contact.

This has led some hoteliers to rethink the potential of robots. To be sure, most hotels do not want cyborgs greeting guests, but robots can be valuable in other ways. And recruiting and retention can be easier if some of the less desirable tasks currently performed by humans can be transferred to robots.

Mobile robots can play a role in addressing this issue. It is not about staffing the front desk with talking robots. While such products exist, the technology as still early stage, not capable enough to provide a satisfactory experience to most hotel guests. The Henn na Hotel in Nagasaki, Japan opened in 2015 with a full staff of 243 robots, but ended up “firing” half of them less than four years later due to complaints from both guests and human staff.

The growth area for robotics in hotels today is where mobile robots serve as an extension of human staff. In hospitality, this means taking menial tasks that have little or no impact on the guest experience and giving them to a robot that works alongside (not instead of) humans. Robots never complain, call in sick, or deviate from standard operating procedures. Used properly, they can be excellent employees, wherever a human touch isn’t necessary or where humans don’t like the work. The term “cobot” has been coined to describe a robot that works in collaboration with a human. In the short to medium term, most robotic applications within hotels will fall in this category.

The technology for autonomous mobile robots has advanced remarkably, thanks to massive investments in self-driving cars by the likes of Tesla, Ford, Google, and Uber. The benefits have spilled over into autonomous service robots, yielding better functionality and lower cost. To be sure, these robots are still not cheap, with unit costs typically in the (low) tens of thousands of dollars. And a robot that might last five years can be similar in cost to paying an unskilled human for just one (although this isn’t a valid comparison, for reasons discussed below). Unlike humans, robots can work 24 hours a day (as long as they get a little charging time), 365 days a year. They don’t incur payroll taxes or fringe benefit costs, and they can perform tasks consistently to quality standards.

Today’s article will focus on just a few of the interesting robotic technologies that have emerged in hospitality.

Types of Robots

Four types of robots are getting early traction in the hotel industry. Some robots can perform multiple roles, while others are single purpose.

Perhaps the most promising category is cleaning robots. Does a customer care if their room was cleaned by a human or robot? How would they even know? Robotic cleaning can be arbitrarily thorough, whereas managing the quality of human work is hard. Robotic vacuums have been around for about a decade, but more sophisticated cleaning robots are emerging today. This one from Peanut Robotics (a picture is worth a thousand words; see the very short video at that link) can do most of the tasks involved in guest bathroom cleaning, for example. The housekeeper removes the dirty linens, wheels in the robot, and starts it. The robot then cleans the bathroom while the housekeeper makes the bed. It can spray one of several cleaning solutions, then scrubs and dries using rags mounted on a rotating head, which it changes autonomously several times per room. Tests indicate that it can do this without mess or spatter.

Other products can vacuum the carpet or floor. This typically does not save much time but can result in better cleaning. Still other products focus on floors or carpets in public spaces or function rooms, where few customers are likely to miss the human touch.

Clearly, there are applications for robotics that have no negative effects on customer satisfaction. And if guests perceive greater cleanliness or see more visible use of cleaning equipment in public spaces, it can be positive.

A second category is disinfecting robots. These are often similar to cleaning robots but may or may not be mobile. Most of them disinfect passively using ultraviolet light. They can be used in public spaces or placed in guest rooms. They can work unattended, which is useful because a full disinfection can take considerable time. Using them to disinfect a guest room that is known to have been virus-contaminated limits human exposure. Based on what science believes about the SARS-COV-2 virus, they may not make a big difference in reducing virus transmission (airborne particles will have mostly settled to the floor long before the robot arrives). But many hotels have found that disinfecting robots provide reassuring visual evidence to guests that safer cleaning protocols are in use – an important marketing benefit.

A third category is autonomous delivery robots. These are typically stationed near the front desk, housekeeping supervisor, or kitchen. They are used to transport items to guests or staff, such as toiletries, towels, food, or dirty dishes.  The staff member loads the items into a compartment, tells the robot where to deliver it, and sends it on its way. The robot navigates hallways and elevators, and for a guest room delivery, it notifies the guest when it arrives. Some delivery robots have also been adapted for moving larger payloads, such as hampers of dirty linens, housekeeping carts, food stocks, and even luggage.

A final category is security robots. These may be similar in design to some delivery robots, but their purpose is to watch for suspicious people or activities, doors left ajar, and the like. Typically, they raise alerts to a control center or security staff, who can view the situation through the robot’s eyes and ears and control its actions remotely, or dispatch humans if needed. Security robots may be trained to recognize faces or license plates, to detect elevated body temperature, to capture 360-degree video and audio, to perform scheduled patrols, and to report physical anomalies such as open doors, standing water, or excess noise. Stationary versions can perform similar tasks at entrances and exits.

Robot Suppliers

Robotics technology is hardly new. Industrial robots have been around for about 60 years, and logistics robots for maybe 30. But it is only in the past few years that robots have started to appear in service industries. I have found maybe a dozen robotics companies that are currently active in the hospitality space; there are certainly many more still under the radar. Most are young, but some have been around for six or seven years now and are gaining traction. Larger companies like LG and Softbank are moving in as well, a sure sign of a growing market.

As with any emerging technology, there are tradeoffs in dealing with large vs. small vendors, and different hotel groups will gravitate one way or the other. Big companies can provide more consistent investment and reliability at scale, with broader partnership and support networks, but they may be slower to innovate and less responsive to smaller customers. If you want to see the full range of both product and delivery/support capabilities, I recommend looking at as many companies as you can, including both large and small. Some will show you the art of the possible and where the bleeding edge is today (and may have things worth piloting); others will give you confidence that they can deliver and support thousands of robots at scale.

In addition to researching numerous products online, I spoke with four companies for additional perspective, including LG (maker of CLOi disinfecting robots, autonomous delivery robots, and mobile information/guidance kiosk robots), Peanut Robotics (with a cleaning robot designed for hotel guestrooms and bathrooms), Savioke (maker of Relay and Relay+ mobile delivery robots), and Techmetics Robotics, which offers a family of autonomous mobile robots with many uses. I thank key leaders at these four companies for sharing their experience and wisdom.

Several other vendors did not respond to requests for interviews, and one (Knightscope, a manufacturer of security robots) completed an Initial Public Offering just last week and was unable to schedule an appropriate executive in time for inclusion in this article.

Pros and Cons

The biggest potential benefit from robots is usually labor savings. However, most robots cannot fully substitute for a human, so comparing the annual cost of a person to a robot is misleading (not to mention a risky human resources strategy!). More commonly, robots work with a human to increase their efficiency. The bathroom-cleaning robot may come with a fixed annual cost (much like a housekeeper’s salary), but it sits idle for much of the time because it follows the housekeeper from room to room and may be in use for only a few minutes per room. A room delivery robot needs a human to select and/or prepare the items to be delivered, to load them into the cargo bin, to tell the robot where to take them, and to keep the cargo compartments clean. As these cases illustrate, robots may not eliminate humans, but can make them more efficient, and offload tasks that humans dislike.

One unionized Las Vegas hotel did a several-month trial of delivery robots, giving the union the choice to keep them or return them after the trial. The union decided to keep them because the staff discovered that they could send the robot on deliveries that would likely be untipped, giving them more time for ones that would. A casual dining restaurant found that wait staff could handle more tables and spend more time talking with (and upselling) customers when assisted by a robot, also earning more tips and reducing the time spent on less desirable tasks.

The right way to evaluate the financial impact of robots that work alongside humans is to compare the efficiency of the humans with and without them. A housekeeping robot that saves five minutes off a 20-minute room cleaning makes the housekeeper 25% more efficient, meaning that they can theoretically clean 28 similar rooms in a seven-hour shift rather than 21. This analysis is oversimplified but illustrates the proper approach. In most applications, robots can hold enough charge to work until they their services are not needed for long enough to recharge. But if not, you may also need to factor in unproductive recharging time.

For many years, the biggest perceived negative to robots was the loss of the “hospitality” experience – guests being warmly welcomed by smiling staff who were there to attend to their every need. To be sure, this has always been important for some guests, and likely always will be. But every guest is different, and even the same guest may view this differently on different stays. In particular, the pandemic has measurably changed attitudes, with many more guests preferring to avoid human interaction, with its higher perceived risk of infection. And staff shortages have forced hotels to think about how to maximize the impact of the staff that they can hire.

It remains to be seen how these attitudes will play out in the post-pandemic world, but there was already a trend towards less human contact even pre-COVID, and no reason to think that it will not continue once COVID is finally in the rear-view mirror.  Indeed, Savioke, which to my knowledge has been in hospitality robotics the longest, said that many hotels that had installed their delivery robots pre-COVID found deliveries per month tripled during the pandemic. In addition to the contactless aspects, many guests may also be quite happy to get their extra towels delivered by a robot that doesn’t demand a tip.

Even if every guest preferred the human touch, many robot tasks are invisible to guests. A housekeeping robot that cleans the bathroom or vacuums the carpet is usually operating with no guest present (or if the guest is there, the housekeeper is as well). A back-of-house robot that moves supplies around, or that brings food from distant prep areas to the main kitchen or dining areas, need never be seen by a guest. Every hotel needs to decide for itself how “visible” robots should be, but most hotels of any size will have at least some applications where robots can work behind the scenes.

Revenue opportunities exist as well. One large hotel that did not permit third-party delivery drivers into its guest-room towers found its lobby being overrun at certain times with delivery drivers waiting for guests to come down to pick up their food. It was reportedly evaluating using a robot to deliver those meals, for a fee. Similarly, robot delivery could extend room-service hours to the middle of the night without requiring additional staff. Revenues from services like these can help offset the robot’s cost.

Considerations in Evaluating Robots

If, like most people, you have never tried to buy an autonomous robot, you don’t know what you don’t know. In researching this article, I spoke with several manufacturers who are selling into the hospitality industry and asked them what they thought buyers should consider when evaluating options and costs. One of my own key takeaways is that while you may have some ideas about how your hotel might use robots, there are many more opportunities that have probably not yet occurred to you. I recommend going into the process with an open mind. Tell the vendors any specific purposes you have but also invite them to suggest other potential uses. Most guest-facing robots have a lot of down time outside of maybe a few peak hours, so finding other uses can help improve the return on investment. Even the vendors themselves said that their customers often surprised them with new applications they had not previously identified.

The four vendors I spoke with have either been established robotics vendors for a while (although not necessarily in hospitality or in North America) or were founded by individuals who have led previously successful robotics companies. I will not compare these companies here; they all have advantages and disadvantages. With any early-stage technology like this, I recommend looking at as many options as you can to better understand the landscape, challenges, and approaches. I will summarize here what I learned from them that can help readers evaluate their options. Not every criterion will apply to every type of robot.

Payload Options: For delivery robots, the first important point of differentiation is how they carry their payload. Some of the early robot models offered only a single compartment to hold whatever was being delivered. Today, there are many more options, sometimes implemented as different bodies on the same robot chassis. Some have multiple open trays or a single closable compartment, while others have three or four separate closable, even heated compartments. One that I saw can grab a large wheeled cart (such as for dirty linens) and push it to where it needs to go; another has a built-in linen hamper as a payload option.

These options can have a big impact on the cost of ownership. A three-drawer delivery robot might be able to deliver three room service breakfasts in one trip during a busy morning, significantly reducing the cost per delivery without requiring additional units. A robot that can work with independent housekeeping carts can shuttle supplies and dirty linen back and forth all day, while one with its own hamper will have to wait in the hallway until it has been filled up, meaning unproductive time. What matters is how efficiently the robot can manage payloads for the intended use, and what proportion of the time it will be moving things around vs. waiting. Of the companies I looked at, Techmetics had by far the largest number of payload options (ten, shown here), many of them relevant to hotels. In a hotel environment, the ability to switch out the payload carriers can potentially enable one robot to serve different needs at different times of the day.

Dispatching: In almost every hotel application, the humans who must load and dispatch a robot are busy dealing with guests at the front desk, preparing food in the kitchen, or otherwise. Look at how easily and quickly the staff can load and dispatch a robot. Also, check out the tools for locating the robot, identifying its current mission, and estimating when it will be available for the next task.

Chain of Custody: Closely related to payload options is the security or chain of custody. Many models come with compartments that will only unlock at their programmed destination. This is critical for room service delivery to prevent enroute food tampering; it may be needed for other payloads as well. However, compartments have the downside of having a finite (and typically small) size. Consider what your typical delivery will include and make sure that the storage capabilities will accommodate it. A robot with three or four compartments can be programmed to use more than one for a single larger delivery, but then it will be limited to fewer deliveries per trip and a smaller-sized maximum payload, so it may make sense to go for a smaller number of bigger compartments. You may also want to consider repackaging room service deliveries to fit into fewer compartments.

Luggage delivery is a particular challenge. While some robots can handle the weight of several bags being delivered to a guest room, trying to fit arbitrary-sized luggage into a fixed-sized, closed compartment is difficult. An open platform or cage is easier to load but presents security risks, and the robot also needs to be aware of any odd-shaped items that may hang past the edge of the cargo platform and therefore limit the robot’s ability to navigate through narrow spaces.

Elevators and Doors. Delivery robots need to use any elevators and to open any doors (fire doors, noise separation doors, back-of-house doors) that may sit between them and a delivery task. For elevators, the most common approach is electronic integration with the elevator control system, although this can be difficult, expensive, or even impossible with older elevator models. Elevator manufacturers and especially local installers and service technicians are likely to still be unfamiliar with robot interfaces. One robotics company reported that some elevator integrations ended up costing almost as much as a robot, and that scheduling qualified service technicians could take many months. Nevertheless, as robots become more common, the landscape is improving, especially for later model elevator systems.

For situations where elevator integration is not feasible, Savioke introduced an upgraded version of its robot (Relay+) that comes with an appendage that is capable of physically pushing elevator buttons (see video). Other manufacturers are looking at similar solutions. The additional hardware will increase the cost of the robot, but also expand the market to buildings with older elevators.

Depending on the robot, elevator integrations can require specific functionality, such as knowing which car is going to arrive at the floor where the robot is waiting. Some robots can sense which door has opened visually but may be challenged to get to it in time, especially in a crowd; this may require lengthening the elevator car dwell-time parameters (which can be done with some elevators but not others). Additionally, interfaces that require Wi-Fi connectivity can be challenging if (as is often the case) the Wi-Fi signal is unreliable inside of the elevator car. And the robot needs to know when it has reached its desired floor, either through the elevator system interface and a reliable network connection, or by other means (such as air pressure or beacons in the elevator lobby).

The robots I saw can only open doors that are electronically controllable. This means that the hotel will need to invest in electronic door openers for any doors that might block delivery routes. These may be controlled wirelessly, or a robot like Savioke’s Relay+ can push a physical door-open button using the same device as it does for elevators. Electronic integration is required for any controlled-access doors.

Guest Notification. Once a robot has arrived outside a guest room with a delivery, it must notify the guest. There are multiple approaches, none of them perfect. If the robot is interfaced with the hotel telephone (PABX) system, then it can ask it to ring the guest with a message that their delivery has arrived. This is perhaps the most reliable solution, but it is dependent on guest phones that work properly and have not been unplugged (hardly a certainty given their increasingly rare frequency of use). A variation is where the robot notifies hotel staff that it is at its destination, but a human places the call. Alternatively, the robot could (via Wi-Fi or through an existing hotel system) send a text message or call, but doing so may bring the robot into the scope of privacy regulations, and may not work for guests without mobile phones or who have turned ringers or notifications off. In high-end hotels with networked doorbells, the robot may be able to ring them through an interface, an ideal solution but rarely practical.  Because of this challenge, one manufacturer was considering offering a device that could physically knock on a guest door!

But none of these solutions is bulletproof. A critical capability is that the robot be able to call for help if it is unable to complete a delivery for any reason, and especially if the guest does not answer the door. Network connectivity and time-sensitive escalation to a person at the hotel become important, especially for food items that may get cold. Hotels need to think carefully about how they can ensure that staff notifications (which may occur only quite infrequently) get noticed and actioned quickly.

Training. Robots need to be taught the floor plan of the hotel, how to ride the elevators, and where to go to recharge when they are not otherwise busy. While this is typically part of the installation process, you will want to understand each vendor’s specific approach and how they will test it. At a minimum this should start with a walkthrough of the hotel by trained installers who can identify any challenges and propose solutions even before the first robot arrives; if modifications to the building or existing systems are needed, you will want to get them done before deployment. Before you sign on the dotted line, it may be worth seeking warranties that the robot will be able to reach every guest room and other area they need to go unassisted, so that the vendor (who understands the limitations of their product) is properly incentivized to identify any challenges up front and price them into their quote (or allow you to get a third-party quote), rather than for you to discover the issues later.

Avoiding People and Obstacles. Robots need to navigate in tight spaces and react to moving things that may be in their way. Test the robot in live situations. The best robots are somewhat human-like in their motion; if they are moving down a hall and see someone coming the other way, they will (like a human) deviate slightly in the direction they think will most likely avoid the person, but then adjust if the human does not react as expected. Children can create special challenges, as they are curious and may test the robot, as seen as in this video of the Techmetics Techi Butler Robot at the Yotel Singapore (make sure to turn your sound on!). Other important tests include navigating through tight spaces, such as past a poorly placed housekeeping cart or around a cluttered back-of-house environment. A robot should be able to squeeze through a tight space if possible, but also to find an alternate route if one exists, or to call for help. One vendor recommended that prospective customers put robots through a “slalom” test to see how quickly and effectively they can navigate around scattered obstacles.

There is an inherent tradeoff between safety and elegance. Some robots start, move, turn, and stop very smoothly, but may take longer to stop in the face of a sudden obstacle. Others may be able to stop on a dime, but their movements may seem herky-jerky. How you evaluate these tradeoffs will depend on how you intend to use the robot, as well as aesthetic preferences. One hotel has used robots to serve pre-poured champagne in function rooms. This would be a good test, since too sudden a stop might tip over the glasses, while too slow could result in a collision with tipsy guests.

Reliability. While robots are generally reliable, there are many issues that can affect their ability to complete a task. Blocked hallways, overcrowded elevators, loss of Wi-Fi signal, and guests not answering their doors are just a few that are faced even by the most reliable robots – and of course there is the possibility that the robot itself fails. The better robots are connected via Wi-Fi and/or cellular connections to support centers, which can diagnose issues remotely or contact the hotel for further intervention. The support centers should also raise an alert if a robot is unreachable.

One of the more common issues is when the robot does not dock precisely enough at its charging station. The tolerance for placement of the robot can be very small, and if not perfect, it may not charge. It is therefore critical that a robot can dock reliably, and also notify the support center or hotel if it is at the dock but cannot detect a charging current.

Battery life and charging time are important variables, but the need depends on the expected use. A room delivery robot is typically used heavily only in the evening and perhaps breakfast hours; it is likely going to be in use for at most a few hours before it can recharge. Other robots may need to work a full shift before their next opportunity to recharge, in which case the recharging may take a few hours.

Service and Support. As reliable as robots may be, they will need occasional service. Many of the companies in the space today are quite small, and their on-the-ground service capabilities are likely to be limited or to involve third parties. Depending on the role that robots will play, you should pay attention both to Service Level Agreements and to the capabilities and reputation of any third-party service organization.

Certifications. In August 2020, Underwriters Laboratories issued UL 3300, the Outline of Investigation for Service, Communication, Information, Education and Entertainment (SCIEE) robots, describing requirements for operational safety around multidirectional mobility, fire and shock hazards, external manipulation, user classes, and use surroundings. The outline prioritizes the safe operation of robots in a variety of environments where people are present. UL 3300 certification should be a consideration in selection of a service robot, but the outline is only 18 months old and to date I am aware of only one product that has been certified. That will change, and some of the others may be close to certification, but it is an issue that merits discussion and consideration.

Personality and Appearance. Robots can be purely functional, or they can be cute, whimsical, or built around a theme associated with a hotel’s personality. For example, one of Techmetics’ customers had their butler robot that was programmed to say “Beam me Up, Scotty!” when it enters an elevator. Equally important, robots can interact in different ways: conversationally, through a touch screen (often with a “face”), or through lights, sounds, and motions. There is no right or wrong here; it depends on what the robot is tasked to do, and how the hotel wants it to behave. Some of the manufacturers have quite a bit of experience in customizing the robot’s look and feel, while others offer a more standardized product.

Pricing. Most of the vendors offer CapEx and OpEx pricing versions, with OpEx either lease-based or “RaaS” (you guessed it, Robot as a Service).  While the vendor pricing models for the robots are straightforward, there can be a lot of hidden costs, so be sure to evaluate the total cost of ownership. Elevator integration is a particularly challenging area; the fact that a vendor has integrated with a particular manufacturer’s system may mean nothing if the hotel has an old elevator unit that either requires an expensive upgrade to be installed by a trained service technician or is so old that it isn’t even upgradable. Before committing, get guaranteed cost quotes on any integrations and building modifications, whether from the robot vendor, the elevator manufacturer, or a third party; they are in a better position to assess the risks than you will be. If you can’t get a guaranteed quote from someone, it’s a red flag.

Depending on application, you may also want integrations with the PABX system, the Property Management System, the Work-Order Management System, or other systems. All of these can fill gaps or improve functionality, but may also add to the cost.


Is it time for your hotel to look at robotics? If you have 100 or more rooms and are facing severe staffing shortages, the answer is yes, you should be looking today. It may or may not yet be the right time to actually buy – that will depend on several variables – but you can explore how you can extend the capabilities of your limited flesh-and-blood staff and start planning for the  future. The field is evolving quickly and will be an important part of your next workforce. It’s time to start thinking about how and when to leverage these new workers!

Douglas Rice

Discover Return On Experience

Three ecosystems — Hospitality & Leisure, Food & Beverage, and Inventory & Procurement — operate independently and together depending on your needs.


Let's Get Digital

7 Questions to Ask Before You Invest in a Hotel Mobile App