New Robots For The IoT

Until recently, remote telepresence has largely been a fixed screen with a video link between participants, and mostly in business setting. But all of that is about to change once the /Internet of Everything begins to take root.

There is a plethora of development going on in mobile robotic telepresence (MRT). Companies such as iRobot, Cisco, Suitable Technologies, Double Robotics, Adept Mobile Robots, VGo Communications, and others are all developing various flavors of mobile telepresence devices.

“In a way, mobile telepresence is kind of a market trying to find itself,” said Dan Kara, practice director for robotics at ABI Research. “For example, the technology exists, but it is still looking for a reason to be.”

There are lots of companies entering the market, but there is yet a strong demand for such devices. However, these companies are jockeying for position in what they consider to be a big growth market. With the potential of the IoT, the vision of MRT devices in both commercial and residential applications is staggering. In many applications, an MRT can be a mobile presence that can act autonomously or step in for a person if desired.

Types of platforms
Mobile telepresence covers a wide swath of applications. Technically Skype, for example, falls into that category because one can walk about with a tablet while using Skype, but the intended definition of involves a lot more. The real meat to mobile telepresence robotics consists of a mobile robotic platform that integrates simple to ultra-sophisticated telepresence systems with remote-control capabilities.

The most common of these configurations is the mobile platform with a video conferencing platform that offers simple, two-way audio and video communications. This type of telepresence mimics social interaction by allowing remote users to communicate with each other with the same capabilities as would occur in a “face-to-face” interaction. Such a device also allows the remote user to experience some of the ambiance of the MRT location.

Stepping up on the sophistication ladder, environmental assessment becomes possible via a variety of integrated sensors. Various levels of physical interaction can be implemented—lights, temperature and appliances, for example—using wireless or IR networks from the MRT. The most sophisticated of these devices include robotic appendages and articulated motive designs that can negotiate stairs, doors, and other adaptive challenges.

However, for the scope of this article, the focus will be on the general category of social interactive segment of MRTs. This is the type of device that has some sort of display, typically a tablet or a PC, and a motive platform that is remotely controlled by the user. It follows preprogrammed routines or can dynamically react to its environment using sensors. The interface at the MRT is usually local, over a local wireless network, but it can include the cloud.

Applications
MRTs come in all sizes and shapes, depending upon the application. The opening graphic shows some of the more common ones, used for personal interaction. But such devices with various configurations are used in commercial applications such as healthcare and warehouses, as well as corporate environments.

One very recent application is the use of such MRTs in hotel services. Starwood Hotels are using a “robot butler” to do things like deliver snacks, towels, and other items. If a guest needs something, they call the front desk. There the staff places the items in the Savioke SaviOne “Botlr” key in the guest’s room number and the MRT will navigate hallways and floors by using Wi-Fi to call an elevator (see Figure 1). The MRT has a screen that the guest can use to communicate with the hotel staff if desired.

Figure 1. SaviOne Robot, Courtesy of Savioki.

True MRTs are an assemblage of high technology. They include an array of sensors to navigate complex environments. They can raise or lower their profile in response to whomever they are interacting with and go from person to person interactively in a “conversational” fashion. Says Kara, “True MRTs use their sensors to gather information, fuse that information, process that information, and finally act on it.”

It’s all about social interaction
There is a nagging question about why we would even want such complex interaction devices. After all, why couldn’t one just walk around with a tablet using Skype, or have a fixed telepresence solution in a conference room?

“Why would you want a mobile platform that adds complexity?” asks Kara. “After all, these devices have to be charged, they can get stuck on rugs or room transitions. They can break down or even run into people or things. So, really, what is the advantage?”

The answer is social interaction. “The goal of these devices is to foster communication and collaboration at a distance, and make it as natural as possible, such as the MRT’s head moving in a way as to acknowledge the points without the body moving,” he says. “Studies have shown that certain ‘human’ types of functions in these devices add a level of ‘presence’ that is missing from simple tablet to table communications or a telepresence room. Other studies have shown that embodied, somewhat humanoid presence increases the ability of individuals to recognize the remote presence as a real person. This makes the communication more intimate and real, and people listen more intently and are able to understand and remember better.”

One of the more successful implementations of this technology is in health care, particularly for patient-doctor consultations. In a hospital, an MRT can sit next to the bed with the doctor at the other end interacting with the patient. With its array of sensors, it can set itself to a comfortable position for interaction with the patient and can move from individual to individual if there are others in the room at the time, including family and nurses. Sensors will automatically adjust for light, sound, ambient noise and find the best positions for optimal interaction.

Other applications
While social interaction is a primary driver for such devices, it is by far, not the only one. One driver is expert consultation. For example, IBM is using such devices for working on airplanes. Such devices can be used to bring in expert capability. So rather than going to the expense of bringing the expert on site, they can collaborate remotely with the same capability as if they were on site. There are tremendous economic incentives for such programs.

Another application is physical security. Advanced MRTs can be used as “security guards” to patrol the premises (see Figure 2). Again, economics plays a role here as well. In this type of application, this removes the “human” element of such functions. MRTs don’t slough off on the job. They don’t need lunches or breaks, and their observation is indisputable. In static environments, such as night warehouse patrol, they can be preprogrammed to a generally fixed routine and monitored at the central station.

They don’t work in all such situations, however. In more dynamic environment (such as the White House, or transportation hubs), where situational reaction is required, they are not quite sophisticated enough to work well because they don’t have the split-second, intuitive decision-making capabilities. But they can still be used for routine patrol in static areas of such parking lots.

Figure 2. MRT guard at Korean prison. Courtesy, ubergizmo.com.

There are other applications as well, in both consumer and commercial venues, and as this technology matures, expect to see MRTs as part of the landscape.

What’s inside
From a high-level flyover, MRTs are not much more than a computer on a mobile platform. But the technology that supports these devices is on the cutting edge of sensors, optics, mobility, and analysis logic. Such high-end MRTs can automatically find and dock to a charging station when they sense their power is running low. They also can automatically synchronize themselves to the user, and the environment — a tremendous benefit when integrating with the IoT.

To be truly autonomous within the IoT, the MRT must be loaded with advanced sensors, three-dimensional, high-resolution optics, servo motors that can enable it to turn on a dime, and precise motor drives that can respond in an instant to the environment. Such equipment is not cheap, and at the top end such MRTs can cost as much as $100,000.

While an iPad or similar tablet will work fine in many applications, high-end MRTs like the IRobot Ava 500 integrate platforms like Cisco’s EX60 telepresence (based upon the H.323/SIP standard), a large (in Ava’s case, 21-inch) high-resolution display that leverages 1080p30 video resolution (for videoconferencing). They also integrate various forms of wired (IP) and wireless (Wi-Fi, Bluetooth, xG cellular) for interconnect. The Ava integrates LIDAR sensors with 3D imagers, and sonar.

On the mobility platform, Ava uses a holonomic drive with integrated object detection/avoidance technology.

Nevertheless, some issues that remain in question. For example, engineers are still working on determining the most conversation-friendly screen sizes for such robots. Big screens are nice for clarity, but they can create a rather eerie impression that a floating head is walking around. “But small screens, can convey a sensation of surveillance,” says Sanford Dickert, an American Robotics consultant. “Small screens also handicap the person speaking through ‘a shrunken head’,” he says.

Power is another issue, obviously. The more technology that is integrated, the more power will be required, to run them. Much progress has been made in the sensor and observation platforms, but the real culprit, at present, is the motive platform. Moving these devices around still requires a substantial power platform. So depending upon the application, the design must balance operating time and rapid recharge technology. In some cases, the MRT can recharge between uses, but in other applications, such as medical, they are on the line 24/7 so power schemes are an integral part of the deployment scheme.

However, most of these devices, with the exception of the elite such as iRobot’s Ava, are still in a rather primitive stage. Integration of all of the technologies in a seamless fashion is both difficult and expensive. To be true autonomous presences, such MRTs need to respond to voice commands, and not be driven by remote control or preprogrammed sequences. Today, those that do have some measure of autonomy (like the iRobot, which has some capability to reroute or adapt to the environment) still require some human assistance, either in real time to learn what the operator wants, or something like simultaneous localization and mapping (SLAM) of the environment.

“Eventually,” says Kara, “autonomous navigation will be the baseline. You won’t get in the door unless your MRT has some measure to autonomous mobility.”

Security
As one might expect, MRTs have a wider security window than a lot of devices, both on and off the IoT. Devices with remote-controlled cameras and microphones, crawling around hospitals, financial institutions, research laboratory and warehouse, and piping everything they see over the Internet, raise eyebrows with security architects. If their communications are via the typical IP environments, then devices such as firewalls can be used to lock down the security of the data streaming to and from its MRTs.

But there are many other vulnerability holes in these devices. Because all of them use Wi-Fi, Bluetooth, and even cellular, those vectors must be secured, as well. As with most IoT/E devices, the experts say that chip-level security should be the baseline. That way secure keys and authentication protocols can be implemented at the lowest level and the data can be encrypted as well, wireless, or wireline.

MRTs and the IoT
MRTs will be a bit different in the IoT, which will be highly integrated via sensors. These sensors can be used to relay information to the MRT, which it can use to determine its environment. “For example,” says Kara, “instead of the MRT having preprogrammed data to tell it where it is, the area that the MRT is navigating will provide sensory data to the MRT.” The MRT will analyze the data and react accordingly.

That is much more of a live environment than exists now. Ultimately, the MRT can be totally operated by what data it is receiving to accomplish the task at hand, without any preprogrammed or innate information. Where it will go, how it will get there will all be ad hoc, as the MRT meanders on its journey to its destination.

Missive
MRT is a very young industry. Only recently has the processing power, power efficiency and component miniaturization begun to approach the kind of metrics that will be required to make MRTs autonomous, intelligent and “self-aware.” Now we are crossing the line between MRTs and true MRTs. However, MRTs are, perhaps, the first stage of true human-resembling entities that will evolve into the MRTs that so many science fiction works have given us.