Figure 1: UWB tag

Call a personal robot to meet you exactly where you are. Let’s say today, it’s an airport. The robot comes towards you using GPS, but soon hones in on your cell phone signal. Your cell phone is now a beacon for the personal robot. It can read angles and distance even when it can’t see you. As the bot gains line of sight, it uses its camera and laser to come to your exact position while avoiding anything in its path. You wait less than two minutes because the bots are always fanning out to cover the grid, and use a variety of online data to anticipate your needs, such as a plane’s arrival. You have a lot of luggage so another bot sidles up alongside ready to tote your stuff. The vehicles form a wagon train that can scale to meet your needs. Why bother to wait for a shuttle bus or rent a car you have to pay to park? No matter what your group size or cargo, the bots can form an appropriate wagon train to take you where you need to go – reading invisible robot sign posts along the way.

Figure 2: Droid phone app showing vehicles and humans

With no dependency on GPS, this robot-readable, tagging technology works the same indoors and outdoors. One grand vision of tagging is alleviating our dependence on cars. Another is creating robot guides and tag-along carriers that might be seen in a variety of big box stores to help customers find things, carry items and escort them out to their cars. Unlike shopping carts, these bots can go from loading purchased goods into the trunk of one customer to quickly find their way to the next customer.. Store managers can purchase kits of robot-readable active position tags that enable staff members to quickly mark critical aspects of the environment, including hazards and no-go areas, merchandise, entrances and exits, bathrooms, charging areas, etc. When merchandise is moved, staff simply moves the tag as well. Across the globe, bots of all shapes and sizes will use the 5D Intelligent Robot Payload to guide people through retail outlets, big box stores, resorts, hotels and airports.

Each business that uses bots can customize the modular base platform for their environment and uses, and can even have the robot reflect the brand. In some instances the end user pays for robot services via their smart phone. In locations such as government buildings, retail stores and resorts, it could be a command wand to check-in at the front desk or pay for an order of goods or services to be shipped electronically. Some people want robots to ride, while other people prefer robots that guide and ask questions. Others want bots to follow behind, carrying merchandise or luggage while generally staying out of the way. But everyone seems to agree that robots that appear on demand and find their own way back are always welcomed!

Figure 3: GM’s EN-V2.0 vehicle

As the use of robots increases, the possibility for truly seamless interconnectivity grows and with it 5D’s ability to monetize the core capability increases. The same active position modules that provide relative position understanding for the bots can be used as plug-and-play add-ons to automobiles, trucks and buses. Since each module can see all the others, they provide an interoperable form of communication and position understanding between cars, robots and humans with smart phones – the ultimate in interconnectivity. The interoperable 5D software is the universal driver and the nexus for how everything connects with the billions of cell phones in the world becoming beacons for robots everywhere.

I don’t know about you, but I would have tremendous difficulty driving or walking through buildings if there weren’t signs expressly for the purpose of helping me navigate. If even humans need help, why should we think that robots do not? Unfortunately for robots, no one has undertaken this effort to help them, which reduces their reliability. We need to help them by creating a robot-readable world. 5D’s approach is to use Ultra-Wide Band RF tags that allow the robot to perceive the heading and distance to each tag. Based on the unique ID and type of the tag the robot can decide to avoid it, pick up the object it is attached to, follow it or simply display it on a map. Using tags that the robot can read and localize from a distance changes the robotics problem, allowing more reactive, reliable behavior and eliminating the dependence on off-board information like GPS. The tags are cost-effective and provide a simple way for humans who know nothing about robots to interact with the system and reconfigure the behavior of the robot. A user may place a follow tag on a car, another robot or a person. A user may create a supply route using tags as breadcrumbs or may tag cargo to be picked up and carried. Also, tags may be used to mark no-go boundaries or specific hazards such as landmines or IEDs.

The US Army has bought into this concept because it needs robots to be reliable in GPS-denied areas. It also realized that the tags provide a simple means for humans to interact with robots. The tags embed into the everyday lives of the user rather than require them to understand a graphical interface or immerse themselves in a map representation. Most importantly, the tags are not limited to line of sight and don’t require the complex, costly and fragile optical range sensors. The tags work in the dark, in high vegetation, dust and fog. Consequently, they are not only useful for robots, but also for humans who, like robots, can’t see through vegetation, fog or around corners. Ultimately, the technology is not only about robots, but about providing ubiquitous positioning for the world. Unlike satellites required for GPS, it doesn’t require global infrastructure. Rather, it allows nearest neighbor relative positioning that can perform even on Mars. For now, we’re using it as a building block for several large scale military projects and we’re extending the technology out into the commercial sector very soon.

This is not to say that we are not interested in understanding biological intelligence. 5D behaviors try to emulate principles we learn from biology. Some examples are the importance of on-board processing, internal models that do not depend on perfect positioning, our ability to focus on changes, motion and anomalies, and the ability of biological entities to use continual feedback to improve behavior. However, to achieve these goals, 5D does not try to copy human physiology or intelligence just like airplanes don’t use flapping wings.

Another approach to creating intelligent robots has been to create information-rich systems that are networked to large amounts of data including terrain information, maps, global positioning and a vast amount of information from satellites and the internet. These robots theoretically have everything they need to think and act in an optimal way. What could be better? Of course there is nothing wrong with trying to use good information, but serious problems arise when the system depends on accurate positioning, reliable communication and up to date maps and terrain information. Much of the last decade has been spent developing unmanned systems that depend on these assumptions and the result is systems that cannot degrade gracefully in the event of failure or imperfect data. There are inherent perils to centralized control. Consider Star Wars Episode One where all the droids stop working once the control ship is eliminated. Is this the model for our factories and transportation systems? Before robots can deliver on their potential, we need to rethink our expectations and ultimately our technical approach.

Rather than focus on human-like intelligence or on a position-based centralized control approach, 5D is focused on creating high-utility reactive behaviors that can be layered through a “Behavior Engine” to accomplish practical tasks. Although we can use them, we don’t depend on GPS, maps, terrain data or databases of prior knowledge. Our hope is that in laying a strong foundation of reliable and responsive “primitive behaviors” we can eventually build towards an emergent behavior capability that is recognized as “intelligent.” The reality is that intelligence is in the eye of the beholder. Here at 5D we are more concerned with effective task completion and a satisfied user than abstract concepts of intelligence. The internal processes will largely be hidden from the user. What will matter most is the users’ ability to predict and understand robot behavior and their willingness to trust and engage with it. This is something that we can and do measure empirically through extensive real world testing and user experimentation. In fact we find that each user group is different.This is one reason why we believe that although the core primitive behaviors can be universal, they must also be reconfigurable and parameterizable to meet different needs.

After many years of developing robot behavior and watching how users interact, it seems to us that it is not only the science that is important but also the art. The human user cares little whether the search pattern or path plan is perfectly optimal. We don’t plan optimally when we vacuum the floor or mow the lawn. It is much more important whether the system moves with grace and appropriate speed. Does the robot accelerate smoothly and can it hug a curve? Can it respond adroitly when something gets in its way? Can it turn around in a tight space or judge whether it can squeeze through a narrow opening? Can it be trusted to remember a dead end when it has found it? Our reactive behaviors provide the ability to address these basic challenges. They are the foundation of our Behavior Engine and 5D has worked hard over the past three years to develop, test and now deploy this core capability across many different robots ranging in size from 25lbs to 4500lbs.

We like working with robots, but in the end, the most important thing at 5D is not robots, but people. The fifth dimension referenced in our name calls out our emphasis on the human element. We believe that robots can save human lives and improve them, but we are also cautious because we know that robots will impact our workforce, how we interact with each other and how we view ourselves. 5D wants people to be happy when they see our robots and we will work hard to ensure that our robots are fun, intuitive and simple. We are not trying to copy humans, but we do think robots will reflect what we value and help us craft the world we want to live in.