Representative End User Clients
Representative Automation Clients
Representative Software Clients
Yokogawa, a Gold Sponsor of ARC Advisory Group’s 24th Annual ARC Industry Forum in Bengaluru, was represented by Dr. Rajeev Joshi, Deputy General Manager. In his presentation, IA2IA: Industrial Automation to Industrial Autonomy, Joshi examined how industrial organizations can gradually introduce AI into existing control environments and move toward increasingly autonomous operations.
Rather than positioning AI as a replacement for established automation, Yokogawa sees industrial autonomy as an evolution of existing measurement and control systems. AI must work alongside plant infrastructure, operate within defined safety limits, and support operators in making better decisions.
Joshi’s presentation can be viewed on YouTube or here:
Why Industrial AI Needs a Different Standard
AI is already widely used in office environments for tasks such as generating content, analyzing data, and supporting project management. Industrial environments, however, have very different requirements.
A mistake in an office application can usually be reviewed and corrected. In a process plant, decisions can affect production, equipment, product quality, and safety. Industrial AI therefore needs to be precise, reliable, and integrated with existing operational systems and safeguards.
Yokogawa’s vision is to make AI a collaborator that complements human expertise rather than replacing it. The objective is to use AI to perceive current operating conditions, anticipate what may happen next, and help optimize operations while allowing people to focus on higher-value decisions.
From Automation to Autonomy
Yokogawa describes this transition through its IA2IA—Industrial Automation to Industrial Autonomy—framework.
Traditional automation relies primarily on predefined logic. PLCs, distributed control systems (DCS), PID controllers, and advanced process control (APC) execute established rules and sequences, with operators intervening when conditions move beyond what those systems were designed to handle.
Industrial autonomy extends this approach by introducing learning and adaptive capabilities that can respond to changing conditions within a secure and bounded operating environment.
Yokogawa does not see this as an immediate jump from automation to fully autonomous plants. Its maturity model progresses through five stages:
Semi-automated → Automated → Semi-autonomous → Autonomous orchestration → Autonomous operations.
Each stage introduces additional capabilities and is intended to deliver measurable operational benefits before organizations move to the next level.
Yokogawa’s IA2IA framework describes a phased progression from semi-automated operations through autonomous orchestration and ultimately autonomous operations
Extending Control with Reinforcement Learning
One of Yokogawa’s technologies for addressing complex process-control challenges is Factorial Kernel Dynamic Policy Programming (FKDPP), a reinforcement-learning algorithm jointly developed by Yokogawa Electric Corporation and the Nara Institute of Science and Technology.
The technology is intended for multivariable and dynamic processes that can remain difficult to control using PID or APC alone. FKDPP uses operational data to develop control policies that can adapt as process conditions change.
The aim is not to replace existing control systems. Instead, autonomous control can be introduced in areas where conventional approaches still require frequent operator judgment, while existing safety and control layers remain in place.
Putting Autonomous Control into Practice
Joshi highlighted two industrial applications of this approach.
At an ENEOS Materials chemical plant in Japan, Yokogawa applied FKDPP to a distillation process that had previously required experienced operators to adjust a valve continuously because PID and APC had not been able to provide sufficient automatic control.
The AI-based controller was introduced after simulation and testing and was used to manage several competing process objectives, including product quality, liquid levels, yield, energy use, and disturbances affecting process temperature.
A more recent example involves the acid gas removal unit at Aramco’s Fadhili Gas Plant.
Yokogawa followed a phased deployment process. Historical operational data was first collected and used for AI training. The control policy was then validated through simulation before AI was introduced into non-critical portions of the process. Operator confidence and safety integration were established before wider autonomous control was deployed.
Yokogawa reports initial results of 10–15 percent reduction in amine and steam usage, approximately 5 percent lower power consumption, improved process stability, and a significant decrease in manual operator intervention.
F
At Aramco’s Fadhili Gas Plant, Yokogawa reports lower amine, steam, and power consumption together with improved process stability and reduced manual intervention following the deployment of autonomous control
Assess, Pilot, and Scale
The broader lesson from these examples is that industrial autonomy should be approached incrementally.
Joshi recommended starting with the operational problem rather than beginning with the question, “Where can we use AI?” Organizations should identify processes where better control could deliver meaningful value, assess whether AI is appropriate, validate the technology through a focused pilot, and scale only after operational performance and operator confidence have been established.
This approach is particularly important in industrial environments, where AI must coexist with established control systems, safety requirements, cybersecurity controls, and experienced personnel.
Industrial autonomy is therefore less about removing people from operations and more about changing where their expertise is applied. As repetitive interventions and complex monitoring tasks become increasingly automated, operators can focus more attention on optimization, exceptions, and higher-value decisions.
For Yokogawa, industrial autonomy is not a distant vision or a disruptive leap; it is a disciplined journey built on trusted automation, proven AI use cases, operator confidence, and measurable business value. The future of process operations will be shaped not by replacing human expertise, but by elevating it through safe, adaptive, and intelligent autonomous systems.