Driving the No-Code Robot Revolution
June 27, 2022
Robotic process automation (RPA) supports mass customization and flexibility and drives the no-code robot revolution. Also called low-code automation, RPA enables programming robot applications and processes by operators and users without reliance on software engineers
Robotic process automation (RPA) and intelligent process automation (IPA) are driving the no-code robot revolution supporting the mass customization and flexibility that are hallmarks of Industry 4.0. RPA, also called no-code or low-code automation, enables the programming of robot applications and processes directly by operators and users, without involvement from software engineers. Add artificial intelligence (AI) and machine learning (ML) to basic RPA, and it becomes IPA, an even more powerful toolset.
Robots were introduced during Industry 3.0 to support mass production and need to be programmed by highly-trained software engineers. RPA represents the next stage in the evolution of robot programming. Initially, RPA was focused on back-office activities and designed to automate repetitive, rules-based, and high-volume tasks such as logging into websites, copying and pasting data from one system to another, and compliance management in healthcare, insurance, finance, and other industries. With the emergence of collaborative robots (cobots) designed to work interactively with people, RPA was expanded to include various manufacturing, warehousing, and logistics activities.
RPA can benefit from the addition of AI, but RPA is separate from AI, which leverages data and experience. The addition of AI can extend RPA beyond its rules-based structure and enable it to deal with some level of ambiguity. The next stage of robot programming, in Industry 5.0, will produce cognitive automation (CA) that will be knowledge-based, behavior-oriented, and not constrained by rules.
This article explores the basics of RPA and its application in software bots, reviews mining processes to program software bots and teaching tools for training cobots, considers the challenges with the identification of the optimal uses for RPA within an organization, looks at the emergence of IPA and how AI is adding flexibility to RPA, and closes by briefly peering into the future and the emergence of CA.
An RPA software bot performs one or more tasks previously completed by humans. RPA can be integrated into legacy systems and support faster and lower-cost digital transformations across an organization. RPA began with simple and routine structured tasks such as web crawling, generating standard forms, and checking forms for completion. The RPA software bots' applications have expanded, and they are now used for customer interactions and revenue-producing processes. RPA software development tools employ simple interfaces such as drag-and-drop, enabling users to automate routine rules-based tasks without the help of a software developer.
RPA can free people to spend more time on higher-value creative and analytical activities. Identifying and prioritizing the best applications for RPA within an organization is not always that simple. Tools like process mining and process discovery were developed to help identify RPA applications with the highest return on investment (ROI) potential.
Mining Processes and Discovering Opportunities
Specialized software bots are used to implement process mining and process discovery. While the goals of mining and discovery are similar, they occur within different parts of the network, are implemented differently, and can be used independently or as complementary activities. Mapping business processes by reading activity logs on computer networks enables process mining to provide the raw data human operators need to make decisions on RPA deployments.
While process mining focuses on network activity logs, process discovery tools run on individual user computers, identifying how specific tasks are implemented. Process discovery tends to be more sophisticated than process mining. It can employ computer vision, neural networks, and AI/ML to analyze human activities and create models of how tasks are performed. The models can be analyzed to identify the best opportunities for deploying RPA bots.
Process discovery brings several benefits:
- Identify nuances in task performance that may be missed using the traditional method of one person shadowing another and taking notes
- Enhance transparency since they are performed with consistency
- Eliminate personal bias and protect privacy
- Speed the deployment of effective software bots
Teaching Replaces Programming
RPA tools enable operators to efficiently train cobots without using advanced programming languages such as C and C++ that are needed to use traditional industrial robots. Operators can use no-code tools such as drag-and-drop apps on tablet computers, pendants, and even manually move the cobot arm from point to point. Cobots are usually lightweight and easily moved from task to task. As they are moved around, they need to learn new tasks quickly. RPA tools enable operators to quickly and efficiently teach cobots tasks such as pick-and-place, process finishing, and quality control.
RPA is especially suited for automating processes in warehouses, logistics centers, and factories with frequent tool changes and/or small batch sizes. A single cobot can be used to perform a variety of processes, improving the ROI of the cobot. That can be especially important in small- to medium-sized operations. The combination of RPA and cobots can eliminate monotonous, time-consuming, and unpleasant process steps and improve worker morale and productivity. However, whether considering software bots or cobots, RPA is not a perfect solution for all applications.
RPA is Not a Panacea
There are no industry standards for RPA. That lack of standardization can be a serious weakness. Without a well-considered RPA strategy, islands of RPA solutions can emerge that will limit the scalability of RPA across an organization and cause the reduction in ROI from using RPA. The lack of interoperability and compatibility of RPA solutions from different vendors means that RPA solutions need to be managed strategically and centralized. RPA is expected to bring scalability as well as agility to an operation. Agility is readily achieved using most RPA tools. Achieving scalability is not as easy.
Process mining and discovery tools for the development of software bots vary from vendor to vendor, and the results are generally not compatible. Accurate and efficient tools are needed to identify the best opportunities for RPA development. Process automation opportunities are specified in different ways by current process discovery tools. Manual intervention is often required to use the results of process discovery tools, increasing the time and cost associated with finding processes that are suitable candidates for RPA software bots.
Similarly, RPA programming tools and methods used with cobots vary from one supplier to another. Further complicating RPA deployments, different RPA platforms can be better suited to specific applications. The needs of early adopters of RPA may not align with the longer-term uses needed to support scalability across the organization.
RPA is designed for use in rule-based activities and is not generally suited to processes or activities that:
- Have frequently changing rules
- Are not completely rule-based
- Are complex with multiple interactions and sub-processes
- Cannot be performed without user intervention and cognitive inputs
When Rules Are Not Enough
RPA is well-suited for developing software bots or programming cobots for specific routine tasks or following processes defined by an end-user. Adding AI/ML to RPA results in IPA bots that can adapt to changing environments. IPA can add capabilities such as exception handling to basic RPA processes.
Giving software bots and cobots AI/ML tools can also imbue them with an ability to handle limited cognitive activities, including:
- Understanding semi-structured or unstructured data
- Visualizing screens such as virtual desktops
- Reading written characters or 2D barcodes
- Recognizing speech and having chats on focused topics
The addition of AI/ML can result in interactive software bots or cobots making rudimentary decisions and interacting with other bots and people. AI/ML are also being employed with process mining and discovery processes to identify new ranges of automation opportunities to scale robust automated networks across organizations.
The development of IPA is another step toward CA. IPA bots are task-focused based on rules, with the addition of some knowledge derived from AI and ML to improve task performance over time. Cyber-physical systems and the IIoT enable the IPA bots of Industry 4.0. IPA bot development prioritizes process automation with minimal human involvement.
Industry 5.0 will extend beyond today's cyber-physical systems and bring people back into the picture. It adds opportunities for complex interactions between machines and humans. CA is expected to emerge in Industry 5.0 and will be knowledge-based and behavior-oriented. CA will move beyond the rules and task limitations of RPA and IPA. CA will mimic human behavior and enable machines to interact with people. Realtime signal processing and AI/ML for speech and image recognition will be among the key differentiators of CA systems.
CA is still a nascent concept to combine tools such as natural language processing, image processing, pattern recognition, and contextual analyses to understand dynamic environments, enabling bots and machines that are self-teaching, can autonomously learn new activities, interact with people naturally, and can modify their behavior in real-time.
RPA and IPA enable users and operators to implement rules-based software bots and cobots without support from software engineers. The uses of RPA and IPA extend from back-office processes to manufacturing, warehouse, and logistics operations. These technologies can free people from routine, repetitive, and mundane tasks and allow them to focus on higher-value activities involving creativity, analysis, and decision making.
RPA and IPA bots are not suited for all activities, and the use of process mining and discovery tools can help identify the best deployment opportunities. The lack of standardization is an important consideration when selecting the best RPA/IPA tools for a given organization. As Industry 5.0 emerges in the future, IPA bots are expected to evolve into CA systems that enable dynamic collaboration between robots and people, further boosting productivity.