Third-Party Integrators and Technology Providers Are Essential to the adoption of Cobots.

Collaborative systems have wide applicability, but initially they were used for light material handling processes such as loading, unloading, placing, or manipulating material for manufacturing operations, such as part transfer and machine tending. Over time, as new robots have come to market and additional software and grippers have been made available, collaborative systems have been employed for additional classes of operations.

Collaborative robots (cobots) are no longer a niche subsection of industrial robotics, having matured into a growing market that is helping businesses achieve significant Return on Investment (ROI).The sales of cobot systems, excluding installation, integration, and programming services, will exceed US$1.23 billion in 2020, up from US$292 million in 2017. ABI Research forecasts that cobot sales will increase to approximately US$13 billion in 2027. Revenue growth is largely a function of unit sales. Approximately 11,000 cobot systems shipped in 2017. This figure is expected to ramp up quickly afterward, reaching more than 630,000 units in 2027.

One perennial problem of the robotics industry that is holding back cobots is upfront Capital Expenditures (CAPEXes) and complexity of implementation. While this problem is alleviated by the lower cost and functionality of collaborative systems, these are still products that have an Average Selling Price (ASP) of over US$26,000 per unit. This does not take into account the cost of installation and building a collaborative robotic solution, or of training staff and organizational change. This is precisely the reason why cobot adopters are mainly automotive and electrical and electronics manufacturers, not the Small and Medium-Sized Enterprises (SMEs) that the industry expected.

Small-to-medium manufacturing is a very sizable sector. For example, it is estimated that over two-thirds of Europeans working in manufacturing are employed in Small and Medium-Sized businesses) (SMBs) (roughly 21 million in 2.4 million companies according to the European Commission Joint Research Centre). According to the U.S. Small Business Administration, approximately 300,000 small-to-medium manufacturers are located in the United States, employing over 8 million workers, and accounting for over 40% of all production value. SMBs play a similar large role among Asian and European manufacturers.

Cobots have been marketed as a solution to this constituency that has struggled to access the benefits of traditional industrial robots. In order to take advantage of this, there has to be increased adoption of Robotics-as-a-Service (RaaS), a business model that provides a way to shift CAPEX to Operating Expenditure (OPEX), and to delegate the difficulties of implementation to experts.

Currently, no cobot manufacturers are touting RaaS as their primary business method, but third-party distributors are. Two of these include Hirebotics and Ready Robotics, both of which focus on providing services related to cobots, albeit in a different manner. Hirebotics is a provider of articulated robotic arm services. Given the very high capital investment required to deploy robotic arms in factories (usually US$100,000+ per unit), there has to be sufficient expertise in the deployment process. Hirebotics provides said expertise but also provides businesses with greater flexibility. Through cloud computing and a mobile app, the company allows customers to hire robotic arms at an hourly rate, offering a high level of flexibility to what would otherwise be a strategic long-term decision with high levels of investment. The company waives any preventative maintenance or upgrade fees and requires no long-term commitment on the customer’s part. Hirebotics does not create its own robots, but generally uses those of the market leader in cobots, Universal Robots.

Ready Roboticsis a third-party integrator that leases out TaskMate, a robotic platform that encompasses a cobot (usually from Universal) and a mobile base for ease of use. This is powered by the company’s operating system Forge, which provides much of the benefits associated with other operating systems (like Rethink Robotic’s Intera 5). These include using simple decision-tree building blocks to create complex solutions, intuitive training, and not requiring any coding or programming experience. 

In the broadest terms, a cognitive system is an architecture that is responsible for the cognitive work of knowing, understanding, planning, deciding, problem solving, analyzing, synthesizing, assessing, judging, and other human capabilities. While no Artificial Intelligence (AI) can replicate human cognition, the term “cognitive systems” is often used as shorthand to describe an architecture AI and Machine Learning (ML) and the capabilities they provide.

A further key component of cognitive systems is that they are defined as anthropocentric, and relate to human capabilities. As a result, high emphasis is placed on vision (machine vision), learning (machine/deep learning), speech recognition (Natural Language Processing [NLP]), etc. In the simplest terms, cognitive systems work by feeding data into an algorithmic architecture, from which AI and ML techniques derive insights and uses without strict programming.

Cobot manufacturers already leverage the benefits of cognitive systems, as highlighted by the development of functional visual programming interfaces that use decision trees to allow operators to train a robot without having to code or be proficient in programming. Some of the most popular use cases for cognitive systems in cobots relate to defect detection, predictive maintenance, and performance-related issues like strength optimization. Through the ingestion of sensory data, Industrial Internet platforms like Tend.ai and Fanuc’s FIELD are providing insights that help cobot operators preempt mechanical issues and detect problems. However, there is demand for more sophisticated solutions that can improve the human-robot interface in relation to more complex environments like high-mix automation. Southie Autonomy Works is a fledgling Boston-based robotics company that hopes to simplify the process of telling a cobot what to do with a wide variety of objects through a hybrid system of machine-vision and augmented reality. Though at an early stage of development, it represents a general trend of third-party solution providers riding on the wave of cobot adoption.

Other ways of potentially improving human-robot interaction include speech and facial recognition. Machine-vision companies like Ever AI and Eyeris have courted partnerships with consumer and commercial robotic vendors that need their platforms to recognize emotions and be receptive to human-voice command. These are not capabilities currently being sought with high priority in collaborative robot contexts, but there is increasing demand for improved simplicity in commanding and delegating to cobots, thus these use cases should not be ruled out. Even companies nominally related to hardware, like gripper companies, are increasingly becoming hardware agnostic and are defining themselves by their ability to leverage AI and ML to improve the range of items they can identify and manipulate, a prime example being RightHand Robotics. On this cutting edge of this development, a collaboration between researchers at UC Berkeley and German automation giant Siemens has developed a gripper system that learns how to grip different objects through a combination of 3D sensors and deep neural networks, in which object images are fed.

Given the complexity and breadth of the wider cognitive system market, and the fact that many use cases remain largely untested or even understood, cobot manufacturers will need to rely on third-party providers for a considerable time to reap the benefits.