A recent trend in manufacturing automation is the use of what have become known as “collaborative robots.” These are robots that are safe to work alongside human workers, as opposed to traditional industrial robots that are generally separated from humans by safety cages. They are also typically easy to program and inexpensive. These properties contrast with those of traditional industrial robots, which are expensive, not safe, and require an expert to program them. The properties of collaborative robots enable new classes of applications that are too low value or too variable to be cost effective with traditional robots.
This paper reviews the economics of automating tasks using collaborative robots and the kinds of new tasks enabled by their use. It describes examples of collaborative robots on the market, and some of the technologies that enable them to be safe, inexpensive, and smart.
COST AND FLEXIBILITY
A common thread in all manufacturing businesses is the desire to improve the efficiency and reduce the cost of the manufacturing process in order to increase margins and thus profits. A common way to do this is via automation, which explains why in the United States manufacturing productivity has increased steadily over the past 70 years while employment in the sector remained roughly constant (Strauss 2014).
But cost is not everything, as in recent years there has been a trend toward smaller batch sizes and more customized manufacturing, driven by consumer demand. For example, automotive manufacturing is set up to achieve economies of
scale by mass-producing a limited range of models. However, this approach makes it difficult to respond not only to the demand for customized features per vehicle but also to the need for different volumes (e.g., demand for hot-selling models as opposed to less popular versions). Manufacturing therefore has to accommodate both cost and flexibility.
There are a variety of approaches to automation with different cost/flexibility tradeoffs, and they are driven somewhat by the properties of the automation technology used. Fixed automation, which uses custom machinery for most or all of a process, tends to be expensive to design and create but very efficient once implemented. It is, however, inflexible and so requires long production runs to justify the expense. Fixed automation is common in industries with stable, long-running production of, for example, consumer packaged goods such as diapers.
Traditional robotics is more flexible than fixed automation, but still has a high cost. Cells running robots are expensive to design and set up, and require long runs to get a return on investment. The dominant market for industrial robots is the automotive sector, where a spot welding robot can be used on a variety of models, yet be tweaked or reprogrammed as necessary as vehicle body parts and shapes change.
The most common automation method uses machinery for high-value parts of the manufacturing process and human labor to complement the machinery. For example, a Computer Numerical Control (CNC) milling machine can be used to turn metal slugs into parts (a high-value operation), while being tended by human operators (who perform the loading and unloading that is of lower value). This is a very common approach because it yields cost savings and flexibility on how a line is constructed and used, but it is more expensive in terms of running costs than fixed or robotic automation.
The technology properties of collaborative robots—safe, inexpensive, and smart—are different from those of fixed or traditional robotic automation, making them more appropriate for low-value and variable processes. The real value of their properties boils down to cost and flexibility. Safety reduces cost, both directly (there is no need to buy an industrial safety system, which are by their nature highly reliable and thus expensive) and indirectly (the floor area taken by safety systems cannot be used for manufacturing). Safety also increases flexibility: the risk assessments required for each application are the same, but there is no need to spend the time and money redesigning and redeploying a safety system for each application. Having inexpensive hardware obviously reduces overall cost, and ease of training reduces application cost, ongoing maintenance, and redeployment costs.
By offering low-cost and flexible automation, collaborative robots are appropriate for use in many areas that are not currently automated (low-value, variable tasks). These include machine tending, kitting (depositing parts into a kit for an assembly operation), line loading and unloading, and packaging, many of which are largely not automated.
EXAMPLES OF COLLABORATIVE ROBOTS
The following sections describe some of the collaborative robots currently available. For a fuller review, see citation Robotiq.
Universal Robotics (www.universal-robots.dk/) sells two collaborative robot arms, the UR5 (with a 5 kg payload) and the UR10 (10 kg payload). These are both six-degree-of-freedom arms, with about 1 m reach. Safety for these robots comes from their low payloads and speeds, and they are inexpensive (around $35,000 for the UR5). The programming interface is very simple and easy to use, allowing quick training and retraining of the robot by users without programming skills. The company also provides support for communication with machines and other pieces of industrial automation.
The Baxter robot, a humanoid robot with two seven-degree-of-freedom arms, is a product of Rethink Robotics (http://rethinkrobotics.com). Its safety is achieved by having arms with a low payload (2 kg) and by using an actuator technology called series elastic actuators, which embeds springs in each joint of the arm, making the arms inherently compliant.
Series elastic actuators, invented at MIT in the 1990s (Pratt and Williamson 1995), consist of a spring in series with the output of an electric motor and gearbox. A sensor measures the twist of the spring, and a control system is used for the output torque at the joint. The spring and control loop enable good performance with inexpensive components, because the spring naturally cleans up some of the undesirable properties of inexpensive gearboxes. In addition, the torque sensing at each joint that this type of actuator affords opens up different strategies for controlling robots, using force control rather than position control.
The use of series elastic actuators allows the cost of Baxter to be low ($30,000), and the robot comes preintegrated with sensors (e.g., force sensing, cameras) that are intended to make the integration process easier. Baxter’s user interface is very different from traditional robot programming: it is programmed by demonstration and consists of manipulating higher-order primitives (picks and places) as opposed to the normal programming method (based on lower-level functionality such as moves). This opens up use of the robot to nonprogrammers.
Precise Automation (www.preciseautomation.com) produces the PF-400, a small SCARA robot with a small reach (0.5 m) and payload (1 kg). Safety is
achieved by its low power and force limiting features. The robot cost is also low (not published but likely under $20,000). The robot programming environment offers a teach-by-demonstration mode for quickly training key points in the robots environment, although it is otherwise trained like an industrial robot.
Market forces and business realities continue to prompt investment in ways to reduce cost and increase flexibility in manufacturing processes. Traditional fixed and robotic automation can offer efficiencies but tends to be inflexible and require large batch sizes to obtain return on investment. There is an opportunity for automation that can be both efficient and inexpensive enough to work on lower-value operations and flexible enough to be repurposed for variable or small batch sizes.
Collaborative robots are one automation choice that meets these needs. New technologies and products enable the development of robots that are safe to be around humans (which in turn has cost and flexibility benefits), inexpensive (the robot hardware is inexpensive), and flexible (their user interfaces are designed to make them easy to train and repurpose). These robots are expected to complement existing automation approaches and provide more opportunities for greater productivity in the manufacturing sector.
Collaborative Robot Ebook, Robotiq. Available from http://blog.robotiq.com/collaborative-robot-ebook.
Pratt GA, Williamson MM. 1995. Series elastic actuators. Proceedings of the 1995 IEEE/RSJ International Conference on Intelligent Robots and Systems, August 5–9, Pittsburgh. pp. 399–406.
Strauss W. 2014. Is the US losing its manufacturing base? Presentation at the Rocky Mountain Economic Summit, Afton, WY, July 10. Available at https://chicagofed.org/digital_assets/others/people/research_resources/strauss_william/07-10-2014-rocky-mountain-economic-symposium.pdf.