A Shift Towards Industry 4.0 Is Improving Manufacturing Efficiency And Increasing Innovation
In Part II of our series with Michael Ger, Managing Director of Manufacturing and Automotive at Cloudera, he looks in greater detail at how AI, big data, and machine learning are impacting connected living and the evolution of autonomous driving.
He also provides key insights into some of the core challenges manufacturing companies face in adopting new Industry 4.0 technology like artificial intelligence, machine learning, and automation.
If you missed the first part of this series, you can catch Part I of Michael’s interview here.
Hi Michael, thank you for joining us again for Part II of our series looking at Industry 4.0 and Artificial Intelligence. To start us off we’ve heard a lot about how connected living, connected mobility, and AI are all big focuses for many heavyweight manufacturing companies, how do you feel Cloudera can support this evolution?
AI enables products, whether it’s robots in a factory, smart appliances within a home or self-driving cars to make dynamic, intelligent decisions.
However, how these devices behave from an AI perspective all starts with data.
Data provides the basis for understanding how the device is used by representative users. Once collected, Machine Learning can be leveraged to predict how other people will use these products under specific conditions and then proactively take actions to serve them better and improve their experiences.
To accomplish the above, Cloudera enables the Edge to AI analytics lifecycle, which includes the collection of large volumes of data and the development and deployment of machine learning models to devices allowing them to respond in a much more dynamic, personalized manner.
In line with the promise of delivering secure, reliable, and intuitive AI-driven solutions, how can Cloudera help meet these criteria?
In today’s complex IT environment, where cloud deployments and data privacy concerns are pervasive, Cloudera can provide critical capabilities to manufacturers in enabling an Edge2AI analytics lifecycle given the plethora of system deployment options and data governance requirements.
Companies across manufacturing industries have implemented differing architectural approaches to deploying the Edge2AI analytics lifecycle. Some companies have deployed on-premise systems for data storage, compute and developing machine learning models, while other companies are doing the same tasks in the cloud. Still, other companies desire to implement hybrid models, managing core data operations on-premise, while providing the ability to “burst” capacity to either a single cloud (i.e Microsoft Azure) or multiple cloud environments (on AWS, Google Cloud Services, Microsoft or any combination of the three).
In addition, these data sets are often very private, so companies must be able to restrict access to critical data sets (i.e. those with personally identifiable information) to specific people.
As previously mentioned, data sets can reside in a potential multitude of locations (On-Premise, Cloud(s) and Hybrid Cloud). To manage this complexity in a centralized and efficient manner, a single pane of glass solution is required to manage the security and governance of this data irrespective of whether it is on-premise or in the cloud. Consequently, data access and governance rules can be centrally defined, managed and tracked, significantly reducing the complexity, cost, and risk of deploying systems worldwide.
The manufacturing industry has many growth areas, for instance, autonomous driving, how do you see this technology impacting the future of the industry?
Teaching a car how to drive is the “mother” of all machine learning use cases, and it is incredibly complex.
Autonomous driving will have a huge impact on society, however, previous predictions for autonomous drive adoption were overly aggressive. Even though we haven’t moved as fast as initial estimates, self-driving cars are nonetheless in our future.
Self-driving cars will impact both people and the industry profoundly. First, it will provide a critical link in enabling transportation as a service, because once vehicles become autonomous, they can be far better utilized by driving themselves continuously within fleets and making individual ownership of vehicles both inefficient and obsolete. As a result, this technology will drive fundamental industry restructuring as automakers potentially transition from being providers of vehicles to becoming fleet operators.
It is also driving the industry to consider the autonomous drive learning lifecycle as a part of its product development process.
It is no longer about providing a simple machine – it is now about creating a smart product or an AI-enabled product – and that requires sophisticated new skills.
What technological progress drives the development of automated vehicles?
Enabling the autonomous drive learning lifecycle begins with collecting data from cars within autonomous driving fleets, which means automakers have to be able to ingest terabytes of data per vehicle, per day.
This quickly becomes a “petabyte scale” data problem, in terms of collecting all of this data (video, radar, and sensor data) and then using this data to teach the vehicle to drive itself. Machine learning models must be developed to train the vehicle to accurately perceive and react to the conditions it is operating under. For example, a vehicle must be able to differentiate between people, snow, intersections, traffic lights, and other cars.
Enabling the autonomous drive learning lifecycle has been underpinned by two important technological advances: the rise of scalable big data platforms, and massively parallelized computing required for machine learning on huge data sets.
The electric vehicle market is growing rapidly, how can Cloudera support the industry to evolve in support of this growth?
Optimizing electric vehicles is also a very data-intensive endeavour. We are working with automakers to be able to collect data, such as “key on” and “key off” information is being collected to understand where vehicles are being used via geo coordinates to help determine the optimal geographic placement of electric vehicle charging stations. Another use case is building predictive models for when the battery is going to fail and optimizing the battery life of vehicles.
What challenges do manufacturers face to effectively adopt new technologies and concepts?
A critical capability required to advance process performance is the ability to master the big data management lifecycle.
Companies need to be able to ingest, store, process and use data for machine learning models and then deploy intelligence to devices and processes to take real-time personalized actions.
Companies are at different levels of maturity along this lifecycle, with some customers just learning how to ingest data into a data lake, while others are using self-service BI and machine learning to operationalize and unlock the value of this data.
Read our latest white paper on Autonomous Driving for more insights about how Cloudera is helping shape the future of driverless vehicles.