The Enduring Risk of Robots in the Public Sphere

One of the most visible side-effects of the COVID-19 pandemic has been the acceleration of automation solutions. Many industries have seen great increases in workforce automation during the pandemic, a trend that already ramping up in previous years and which has augmented greatly as a result of the pandemic, especially in robotics. A less well-known effect has been some of the conditions for research and development that lockdowns have brought about. All of a sudden, lockdowns presented deserted cities all over the world, a specific set of conditions offered a great opportunity for the planning and development of technologies such as self-driving vehicles and the use of mobile robots in public spaces. Indeed, one of the problems that the deployment of robots in public spaces have encountered in the past is that these technologies struggle with heavily unstructured environments. Mobile robots do well in grid-like structures, with mostly flat surfaces and very few obstacles around, which are ideally not unknown and unrecognizable. However, most cities in the world (not to mention the countryside and small towns) are not like that at all. But most cities are easier to navigate if empty, and this has spurred some interesting interdisciplinary work. Investigating the Impact of Robots on Public Space, from Melbourne, Australia, a city that has undergone some of the longer lockdowns in the world, which allowed for examining urban robotics in the age of lockdowns, with important repercussions for both the use of technology in public spaces as well as for policy design.   

Advances in technology have allowed robots to become more autonomous and efficacious than ever before, making them increasingly present in urban public spaces, with applications for delivery and cleaning already available in many city environments around the world. The COVID-19 pandemic has encouraged further use of robots in settings ranging from hospitals and urban parks to food delivery and sanitation. This has been primarily driven by the need for infection control, but from the perspective of robotics, the pandemic has brought a simplicity to the urban environment, providing conditions in which robots work best. Cities with more certainty and less contingency are amenable to the way robots treat space, namely, the Melbourne study states, as predictable, partitioned, and datafied environments, and this has certain implications for cities as the use of robotic technologies grows. The question is whether robotic technology can master unstructured environments, and one typical answer is that it is possible through the use of technology. Boston Dynamics’s Spot, a robot that runs on four legs and has the appearance of a dog, has been designed to operate in unstructured environments inaccessible to other robots. During 2020, Singapore deployed Spot in parks across the city to ensure that park visitors were practicing social distancing. Using remote navigation and equipped with a camera and a pre-recorded message, Spot would bark a warning whenever it came across someone not complying with safe distancing rules. In addition, Singapore also deployed a fleet of drones to observe parks from above and measure how many visitors there were at each location. These data were made available on a website so that people could determine which parks had the lowest number of visitors.

The required technology for these endeavors involves the kind of things ABI Research has gone on about for some time now, and which is forecasted to continue to develop at pace. Specifically, SLAM capabilities as well as 2D and 3D LiDAR sensors and cameras are required for 360-degree 3D data, one of the holy grails for mobile robots to be able to navigate unstructured environments (another essential technology involves having enough information about all objects in a given space). But however these technologies may advance, when it comes to the interaction between humans and robots, as ABI Research’s report Bringing Functional Safety and Risk Prevention to Industrial and Commercial Robots has pointed out recently, public policy and safety standards (and the interrelationships between the two) very often prove to be rather important drivers of how the technology evolves. This is usually because policy and standards impose the conditions under which robots ought to behave around people, which imposes what sort of solutions technology needs to meet. This is going to be even more important in public spaces, where robots are likely to encounter all sorts of unknown quantities, which can result in hazardous situations; in one well-known case, a robot failed to recognize a wheelchair user, which is naturally an intolerable situation.

Given the recent trend to further extend the use of robots in public spaces, there may be the temptation to simplify the surroundings that robots are placed in to make it easier for them to navigate and be effective. After all, robots thrive in controlled conditions, and in this sense, a COVID-19 world works to their advantage. Cities designed for robots would be more predictable, navigable, and amenable to datafication, with knowable objects and patterns, as the Melbourne study shows. There are examples of robot-centric spaces, as the implementation of dedicated lanes on highways for self-driving vehicles, in addition to recent policy proposals for dedicated roads solely for these technologies—an absolute necessity, as it is very unlikely that current technology, including Deep Learning models of Artificial Intelligence (AI), will solve the problem of self-driving vehicles, contrary to some of the claims in the market. Well-designed, anticipatory public policy can do much to ensure that society reaps the greatest benefits from new technology while reducing possible harms, a kind of policymaking that has been of great important to robotics.

Unfortunately, in the case of the deployment of robots in the public sphere, there is very little guidance from public policy and international standards to be had, and this will impend development. There are standards regarding “smart cities”, but these do not cover the use and deployment of robots in the public sphere. And what there is available, for instance, the EN ISO 13482:2014 standard, is significantly lacking in detail in this respect, putting pedestrians and bystanders under considerable danger. Indeed, the public sphere showcases a number of problematic issues to solve for robots, most notably crowd density and the need to adhere to social norms (and many more), and it is unclear how to approach these situations. In the past, many robotic vendors have been directly involved in the development of standards and have advice public policy, but ABI Research believes that a much more synchronous relationship is needed. Not a case in which vendors are pushing for their own solutions to be reflected or favor in policy and standards, but a collaborative endeavor to come to an understanding as to how to best deploy robots ‘out there’, and in what circumstances. Robotic vendors have taken great strides into making mobile robots much more autonomous in movement in recent years, but these developments have only been properly cashed out in rather circumscribed settings and environments—is it advisable to suggest that such a situation be mirrored outside the warehouse, too? Something will have to give in the long term; but in the long term we are all dead, as Keynes liked to say, and in the meantime public policy will probably have to establish that unsafe robots be compartmentalized, and instead given their own space next to humans.