Challenging the laws of robotics

This article originally appeared in the July'23 magazine issue of Electronic Specifier Design – see ES's Magazine Archives for more featured publications.

Robotics implementation has accelerated over the past few years, especially in the US, where demand is growing with increasing investment in automation. According to the Association for Advancing Automation (A3), total orders for robots in the US in Q1 2021 were 20% greater than the corresponding quarter in the previous year. Outside of automotive – the industry generally accepted to be at the forefront of robot integration – key industries significantly ramped up their spend on robots: metal processing was up by 86%; life sciences and pharmaceuticals up by 72%; food and consumer goods up by 32%; while the spend for the rest of the non-automotive industries was up by 12% on aggregate. In addition, 2020 was the first year in which companies outside the automotive sector spent more on robotics than those within it.

A major driving factor for the greater introduction of robotics into the wider industrial environment has been Industry 4.0. Introduced by the German government in 2011, the strategy is presented as a concept to improve manufacturing by connecting ‘physical’ production lines to the ‘digital’ of work scheduling, equipment monitoring, statistical quality control, and predictive maintenance. The idea being that factories will evolve into ‘smart factories’, where every physical action on production lines is matched with real-time data capture, advanced analytics, and action requests handled by Cloud-based computers accessed via robust communications networks.

However, significant challenges exist – many of which were laid out in research published by the Interactive Robotics Group at MIT in November 2020.

Issues

A major point is integration into production lines. In the first instance, these often large, fast, and powerful robotic machines need to be isolated to protect factory workers. Secondly, there is the programming issue: the industry lacks a common language for programming robot movement. Each manufacturer usually has its own user interface and different manual controllers. Programming has become a specialist skill often outsourced to third-party integrators, whose work can cost more than the robot itself. And if there is a requirement to change a programmed function because a part has changed, for example, it may become necessary to recall the integrators to make the adjustment.

Even those that have embraced industrial robots may be sceptical about other parts of the ‘smart factory’ paradigm. The Industrial Internet of Things (IIoT) raises issues such as security, privacy, and the control of data. Those facilities using the IIoT to monitor highly automated production lines need to implement extremely robust communications to reach all the distributed sensors and actuators. For example, grocery delivery company Ocado installed a private 4G network to provide secure communications for its very large robotic warehouses.

Then there is the use of Cloud computing to gather, store, clean, and analyse data. The promise is that it provides processing power on demand, but it also needs to be a mission-critical system when it is used to manage a production line. Another major concern is the conversation about ‘information’: gathering a lot of data is not the same as creating actionable insights, and therefore return on investment will be looked at closely when implementing infrastructure to integrate a standalone robot into a smart factory.

Robots in disguise

It is important to discuss a couple of subspecies of robots, namely: collaborative robots, or cobots; and autonomous guided vehicles (AGVs).

The first of these, cobots, enable closer working between people and robots and tend to be smaller and lighter robot arms that are designed to handle lighter objects and move more slowly than standalone robots. The arms will usually include devices such as passive contact sensors, light or laser curtains, proximity sensors, or even capacitive ‘skins’, to send a signal to stop movement if they encounter an obstruction – such as part of a human.

As an example, the Bosch Rexroth APAS Production Assistant cobots can sense people within its working area without making contact and will slow down accordingly or even stop operation. An ISO standard – ISO/TS 15066 – has been developed to define safety requirements for collaborative robots.

The second, AGVs navigate factory floors moving materials under their own control and using sensors to avoid obstacles. 

The development of self-driving cars is helping with key concepts such as simultaneous location and mapping, as well as encouraging the development of more sophisticated sensing technologies such as time-of-flight sensors and LIDAR. However, a lack of standards is making it more difficult to manage a fleet of AGVs from multiple manufacturers.

Enablers

The development of standards for the robotic arena is likely to be one of the key enablers for its future. There is a need for a common approach to issues such as programming, communications, and co-working. Fortunately, there are efforts already ongoing, such as the open-source Robot Operating System (ROS), which is a collection of tools, libraries, and conventions, popular with individuals and academics, to create complex and robust robot behaviours across a multitude of robotic platforms. According to the ROS: “…creating truly robust, general-purpose robot software is hard. From the robot's perspective, problems that seem trivial to humans often vary wildly between instances of tasks and environments. Dealing with these variations is so hard that no single individual, laboratory, or institution can hope to do it on their own. As a result, ROS was built from the ground up to encourage collaborative robotics software development.”

Another key enabler is the development of sensor technology – from simple devices that monitor an AGV’s battery status to advanced devices that help position a robot’s manipulator to pick up a component. These sensors will need to have a series of abilities and will need to be robust, highly accurate, offer long operating lifetimes, and deliver consistency in rapidly changing environmental conditions. For example, the authors of the MIT report found that some sophisticated robot-vision systems worked well under lab conditions but failed in the environment of a real production line with the significant ambient light variations. One option as a workaround for this problem could be the use of infrared time-of-flight sensors, such as OMRON EMC’s B5L sensor, to measure distances to objects and create 3D models of their position in space. The device is designed to reject the effects of varying ambient light and is protected against mutual interference allowing up to 17 of the units to share a working environment. But, however technologically advanced they may be, sensors will need to be backed up by a communications, computing, and decision-making infrastructure that can absorb, analyse, and act fast enough to aid robotic operations.

Future

In the automotive sector in particular, industrial robotics has delivered successful applications over many years, even decades. But major challenges remain if it is to extend its effectiveness and viability across new industrial applications, including some of the same issues it has always faced such as understanding the position of objects in space and the programming required to manipulate those objects at a distance. But undoubtedly, the pathways and strategies exist and are evolving to deliver a higher robotic penetration in the new smart industrial reality.