The 21st century has brought explosive growth in computing power, communication technologies and artificial intelligence (AI) — capabilities that together have delivered what is known as cyber-physical systems (CPS). CPS capabilities include autonomous decision-making, and communication with both machines and humans. These real-time abilities help CPS fine-tune their courses of action and operate in dynamic situations with high reliability. They are tightly tied to the concepts of sensor networks and robotics loaded with computational intelligence.
Consider the potential for CPS in these applications and others:
- Weed and disease detection.
- Hazard detection and mitigation.
- Remote machinery operation.
- Healthcare: tele-surgery and nano-robots.
- Search-and-rescue operations.
CPS and Industry 4.0
CPS is a key component of the fourth industrial revolution, often called Industry 4.0, which posits that game-changing opportunities will result from the emergence of advanced automation and data exchange leveraging the Internet of Things (IoT), cloud computing, data analytics, AI, additive manufacturing and, of course, CPS. With Industry 4.0, the internet, not the computer, is the core technology.
The idea is to digitize all aspects of manufacturing and interlink all machines, processes and people spanning the entire value chain, connecting suppliers, customers and partners into a digital ecosystem. Industry 4.0 could change all facets of manufacturing, and thereby take productivity, efficiency and safety to the next level. In an Industry 4.0 world, production environments are self-configuring, self-optimizing and self-adjusting to real-time operating parameters. This will drive greater flexibility, agility, safety, cost-effectiveness and productivity improvements.
Challenges and potential solutions
For all its opportunity, CPS deployment undeniably poses a distinct set of challenges:
- Adaptation. This is a two-fold challenge: one is to make all assets compliant, and the other is to onboard a CPS to the ecosystem. Seamless machine-to-machine communication is the key to success for any CPS that is added to an ecosystem. Legacy systems typically do not have data interfaces, and when they do, they are rudimentary and store information in a proprietary format. Effective adaptation can be achieved by selecting a universal and standardized data-exchange specification (ISO, for example), which can then be used to map all data fields from legacy assets to the CPS. Starting with a big-bang approach — migrating the existing asset ecosystem all at once — can be overwhelming due to its complexity, cost and operational risks. To overcome this, we recommend an incremental approach. To start the process, identify a business case with high potential for improvement. Once the assets are identified for the given use case, a pilot project should be planned that includes onboarding a CPS that can interact with existing assets.
- Culture. All organizations face a cultural challenge in transitioning to a human-machine based ecosystem in which machines play a prominent role in decision-making. Technology has long brought fears of job loss and redundancy; the advent of CPS is no different. In response, managers must reinforce the value of people in the digital ecosystem, assure them that machines are not a threat but rather extensions of their capabilities, and define the collaboration protocols between people and machines. To promote a digital culture, decision-makers and stakeholders need to clearly understand the changing environment, know the current state of their operations and promote a culture of innovation. They must embrace the benefits of CPS and consider how the technology will align to their long-term strategy.
- Security. Cybersecurity is one of the toughest challenges from an implementation standpoint. As machines and processes become more connected, they are also exposed to vulnerabilities and pose a potential threat to critical industrial infrastructure. CPS security vulnerabilities can bring catastrophic effects, there’s no way around it. With the integration of IoT and CPS devices into existing enterprise networks, it is important to ensure that existing systems are programmed to accommodate the expected benefits and that the technology does not create a significant liability risk. As devices will now be connected outside organizational silos, the exposed attack surface and the associated risk for a cyber-incident increase greatly. Securing against such vulnerabilities involves building middleware that facilitates secure integration of legacy and IoT systems in enterprise networks.
- Talent and skill development. As CPS becomes mainstream, the shift will produce opportunities in the form of new offerings, business/service models, etc. It will also automate jobs that are currently handled manually. If not managed well, it might lead to social inequality where low-skilled people are sidelined by intelligent CPS systems. The key is to strike a balance between finding and allocating value-added tasks to humans and repetitive jobs taken on by CPS. Establish a re-skilling strategy for the existing workforce so they can use their core skills and learn new ones to stay relevant.
Maximize your chances of success
Organizations that move to implement CPS for their operations should consider the following points to help maximize benefits and avoid pitfalls:
- Create strong ecosystem mapping and monitoring. In line with the target operating model, organizations need a strong ecosystem mapping and monitoring framework to help identify and monitor all IoT devices and sensors, data flows, data management processes and various data protocols for devices to communicate. Instating robust ecosystem monitoring will ensure optimal CPS performance.
- Enact use case-driven rather than big-bang deployments. Organizations must thoroughly analyze and prioritize beneficial use cases and trial-test them incrementally. This is preferable to a big-bang approach, as noted above, because associated challenges and risks can be carefully mitigated. It’s important to build and maintain a CPS roadmap over time, starting with less complex implementations and graduating to higher complexities and AI-driven systems.
- Apply modeling, simulation and piloting discipline. After organizations complete planning and design, they should model and simulate CPS functioning by running pilots in controlled environments. As these systems become more autonomous and work in swarms (see Figure 1), challenges concerning the collaborative aspects of different systems should be identified early. This will reduce risk and uncertainty during actual deployments.
- Build a scalable architecture with guaranteed performance. A consistent architecture, scalable at the meta level as well as the physical device level, is important when adapting to CPS-induced changes. This approach offers a loose coupling of the cyber and physical parts to ensure that maintenance is more easily accommodated, while the software base (control loops of the CPS) with performance guarantees is maintained at all times.
- Design for human interventions and overrides. When designing CPS, build in human interventions and overrides. This will help in situations where the self-controlled environments do not function as planned. Also, embed human interventions and overrides to provide a sense of trustworthiness.
- Consider robustness, safety, security and privacy. A critical aspect of the CPS is robustness against uncertainties in the operational environment, security attacks and malfunctions. Deployments must guarantee safety to humans at all times. It’s also critical for CPS implementations to include privacy controls established in compliance with applicable regulations, such as General Data Protection Regulation (GDPR).