Overcoming the Robotics Tower of Babel – Robotics Business Review

Listen to this article

Today, the range of opportunities for robots and other autonomous machines is enormous. The COVID pandemic has seen robots deployed for applications as diverse as UV cleaning of hospitals to last-mile delivery, for logistics and warehouse work, as well as for inspection services for off-shore wind farms. Robots, as well as a growing number of consumer electronics products, are demonstrating increasingly sophisticated autonomous skills especially in the crucial area of simultaneous location and mapping (SLAM), the process of building and updating a map of an operational environment, while simultaneously maintaining the location of a systems location within it.

Where Am I?To operate effectively and safely in dynamic environments among people and other devices, robots must be able to calculate exactly where they are at all times. Using SLAM technologies and techniques, robotics systems must be able to accurately, reliably and consistently answer the question Where am I?, without recourse to external systems like GPS or beacons and other way-point systems.

Bespoke SolutionsMany robotics firms are already creating robots that can determine their position with high levels of accuracy. But each sees the world around it in its own way, and in a manner completely incomprehensible to other machines or humans. Hardware and software set-ups are tailored for specific use cases and operational environments, and the systems will fail if the systems are used in any way other than what they were precisely engineered for.

To illustrate, consider an automated cleaning robot and a hospitality robot working in the same shopping mall. The two robots perceive their operational environment around them completely differently. Each operates in a narrowly defined spatial silo engineered for specific routes, functions and parameters. They are unable to collaborate with other robots, machines, or people.

The tight integration of sensors and SLAM software in each of the cleaning and hospitality robots means that mapping, localization and navigation information cannot be shared between them, and positioning related algorithms cannot be reused in other systems. Without its own bespoke combination of sensors and software, each robot is unable to answer that core question Where am I?

For both start-ups and larger established players, what is required for robotics systems to reach their full potential is access to a common, shared approach for mapping, positioning and understanding their operational environment.

Missed OpportunitiesDeveloping SLAM systems is resource intensive, technically challenging and expensive. In addition, the lack of common and shareable approaches for understanding the physical space around autonomous devices has resulted in a multitude of system specific, spatial intelligence silos. This robotics Tower of Babel threatens the projected growth and viability of the robotics industry, and inhibits the development of devices and robots that could address some of the most pressing challenges in the economy, the environment and society.

Reinvention QuagmireHundreds of innovative, entrepreneurial companies are desperate to deploy their proof-of-concept designs in the real world. These are the next generation of businesses that will drive the growth and increase the value of the robotics market.

Many of these firms have brilliant designs and applications that could literally change the world. But many are stuck reinventing the core technology of spatial intelligence. Each pursues their own approach, establishing new silos, and then struggling to adapt to the thousands of edge-cases that cause their designs to fail in unexpected ways.

For both start-ups and larger established players, what is required for robotics systems to reach their full potential is access to a common, shared approach for mapping, positioning and understanding their operational environment.

Multiple Designs, One ApproachThere is no one-size fits all for robotics systems. They each have their own form factors, hardware/software set ups, and parameters tailored to meet specific requirements. But there can be a consistent and repeatable way of perceiving the world around these devices. Nature provides a clue.

There are thousands of different designs for living creatures on Earth, but a surprisingly consistent way lifeforms position themselves in the world around them. This approach, honed through millions of years of evolution, emphasizes a combination of two eyes and inertial sensors in the inner ear to calculate position.

At SLAMcore, we have mirrored nature by creating a common approach to SLAM using two cameras and an inertial measurement unit (IMU). With these core components SLAMcore is creating a universal language of spatial intelligence that can benefit all robot developers. Simple, effective and widely available sensors, combined with SLAMcore algorithm, provides a consistent, repeatable and sharable way for any robot to perceive, map and describe the world around it.

Using a common language to describe the world around robots also means that information can be aggregated and shared. As such, it can unleash a new wave of growth in the robotics sector.

Vision is KeyEven basic, low-cost standard-definition cameras capture huge amounts of data. Processed in the right way, this information can support instant and accurate calculations of position even with no prior knowledge of the location or physical situation.

The algorithms developed by SLAMcore engineers are able to take visual data to create sparse point-clouds of the features in any scene that a robot or autonomous device can use to accurately and robustly calculate its position. The same data is also used to create detailed 2.5D and 3D maps that add more functionality, including identification of free space that is safe to occupy. Additionally, the software identifies and labels all the objects in a scene attaching semantic understanding of what the robot is seeing. This information is the basis for decisions on how it should react to objects in its environment.

A Common LanguageUsing vision as the primary input for mapping, localization, and navigation creates a common framework a language of spatial intelligence that can be shared with other devices, and with humans. If robots of all types, and the humans that work with and around them, all see physical space in the same way, it is much easier to begin to collaborate, share and build a common understanding.

Using a common language to describe the world around robots also means that information can be aggregated and shared. As such, it can unleash a new wave of growth in the robotics sector. Customers benefit not only from a shared language for spatial intelligence that will short-cut their own development cycles, but it also provides a constantly growing knowledge base.

Just as children learn through trial-and-error, robots, too, can learn from their failures, as well as the failures of other robotics systems.

Constant EvolutionAlthough robots and autonomous devices come in many different shapes and sizes, they tend to fail in common ways. Outside of the lab, robots quickly encounter unexpected situations unforeseen objects, different lighting conditions, new layouts or physical environments. These edge cases are difficult to anticipate, simulate and program for, and they are usually the source of robot failure in the real world. By describing these unexpected situations in a common way using data from cameras and IMUs processed consistently by SLAM algorithms every edge-case, and its solution, can be recorded and shared.

Just as children learn through trial-and-error, robots, too, can learn from their failures, as well as the failures of other robotics systems. In terms of positioning, sharing data and reusable maps allows knowledge about a physical environments to persist so that robots can learn from experiences of other robotic systems.

Virtuous CircleBy analyzing data from hundreds of previous positioning edge-cases, SLAMcore engineers have identified how and why SLAM estimations have failed in the past, and how they can be overcome. They have also developed a consistent frame of reference that allows them to leverage and contribute to a global body of data on how robots locate and map in real-world situations.

As data granularity and scale increases, and more edge-case solutions are found, SLAM algorithms are tweaked to make better positioning judgement calls. As mapping techniques improve, so do robot operations, allowing robotics systems to be used more widely, resulting in the collection of even greater amounts of edge-case data. This process creates a virtuous circle that benefits all participants.

Shared MapsConstantly updated, shared maps have immediate and obvious applications for wide scale robotics deployments. Fleets of robots can contribute to shared maps so that every change is noticed and passed on to all those affected. Robots learn from their peers experiences, and maps can be shared with humans so that both they and the robots have a commonly agreed map that accurately represents their shared operational space, and can be verified with the world around them.

As a universal language of spatial intelligence, digital twins will unleash a wave of robotic innovation and deployment, delivering significant benefits across industrial, commercial and consumer sectors.

Digital TwinsThe accuracy, currency and robustness of these maps, constantly updated as every robot surveys a scene, become the foundation for a real-time digital twin of the physical world. This provides many benefits for those managing facilities, locations or properties, providing real-time data on the exact state of the physical world, but also the opportunity to test and simulate actions before implementing them.

As a universal language of spatial intelligence, digital twins will unleash a wave of robotic innovation and deployment, delivering significant benefits across industrial, commercial and consumer sectors. The value and the utility of a hyper-accurate, real-time digital twin of the physical world, automatically updated and shared as every robot or autonomous device passes through an operational environment, is immense.

Common LanguageAt SLAMcore, we have developed a common language for spatial intelligence allows developers and designers to learn from the errors of others, enabling them to progress further, faster. With tens of thousands of sessions, thousands of hours of operation, and well over three million meters traveled and recorded by our customers, this common language has incorporated deeper knowledge, across a wider range of scenarios, than any individual engineer can hope to master. Moreover, an event or failure experienced by one class of robotics systems (say, a drone designed for delivery), can provide valuable information to a designer of another type of system (a wheeled robot for hospitality, for example).

Tower of BabelThe story of the Tower of Babel was a warning to humankind not to attempt to reach the heavens. But in contrast to the humans in the origin myth, we want robots to share and benefit from a common language, one with which to describe the physical environment around them. We want robots to cooperate with each other, and to work alongside humans to find solutions to some of the worlds most pressing challenges. We want, and many would argue, need, robots to help build a better world.

At SLAMcore, this is the mission that drives us to make quality spatial intelligence accessible to all. If robots and other intelligent systems use a consistent language for shared spatial intelligence, we can democratize access to robust, accurate and fast SLAM. This would open up many opportunities and create the necessary building blocks for an explosion of robotics solutions to address critical challenges such as addressing climate change, supporting pandemic and disaster response, care for the elderly and disabled, as well as providing cost effective and efficient ways to deliver the economics of abundance.

About the Author

Owen Nicholson is the CEO/co-founder of SLAMcore, and an expert in IP management. He specializes in technology commercialization, with industrial experience of early stage technology transfer, intellectual property negotiations, and the successful delivery of multi-million pound projects. In his previous role as Operations Manager at the Dyson Robotics Lab (Imperial College London), Nicholson was responsible for the management of the laboratorys intellectual property and monitoring. He was also Head of Growth at IN-PART, an online platform showcasing academic advances within engineering and life science disciplines and creating opportunities for university-industry collaboration, including co-development, licensing, and long-term partnering, with a focus on technology transfer from the research laboratory to the market. Nicholson began his career as Graduate Scientist and Technology Manager at the UKs Ministry of Defence where he worked at the Counter Terrorism Science and Technology Centre. Holds a BS in Physics and an MS in Innovation and Technology Management from the University of Bath

Originally posted here:

Overcoming the Robotics Tower of Babel - Robotics Business Review