Renesas Electronics/Reality AI

AI - step by step becoming a full-service provider

28. Juli 2022, 17:36 Uhr | Iris Stroh
Dr. Sailesh Chittipeddi von Renesas
Dr. Sailesh Chittipeddi, Executive VP and GM of Renesas' IIoT Group: »AI is definitely a very crucial area. We will therefore continue to expand our ecosystem. Even though we already have hundreds of partners today, there are still areas that we cannot currently serve. This means that we will expand our ecosystem to thousands of partners.«
© Renesas

Renesas Electronics announced the completion of its acquisition of Reality Analytics Inc (Reality AI), effective July 19, 2022. Dr. Sailesh Chittipeddi, Executive VP and GM of Renesas' IIoT Group, in an interview with Markt & Technik, explains Renesas' goals in the AI space.

Markt & Technik: Renesas has acquired Reality AI, a specialist in embedded AI and TinyML solutions for advanced non-visual sensing in automotive, industrial and commercial products. What benefits do you expect from this?

Sailesh Chittipeddi: The acquisition allows us to offer a more comprehensive solution for industrial customers. Reality AI provides the inference models, we provide everything else, i.e. the libraries, other tools, and the workflow for integration into MCUs and MPUs.

Does this mean that Renesas will integrate Reality AI's tools into its own tools?

Exactly. We want to offer a complete transparent workflow for our customers in some markets where it makes sense. That will take a little time, of course, but we're working on it.

Which product groups does Renesas want to support with the Reality AI tools, just MCUs and MPUs?

Yes, primarily our MCU and our MPU families. Reality AI focuses on two segments today, though that's not to say that won't change in the future. But as of today, Reality AI is currently primarily focused on industrial applications, with a little bit of automotive applications on the side. Accordingly, our next step will be to implement the possibilities that Reality AI offers us for the very MCUs that were designed precisely for these applications. If Reality AI will address other markets in the future, we will be able to offer inference tools for these applications as well.

What applications are you thinking of?

Today, for example, Reality AI does not yet have any approaches for vision based processing. But over time, they will certainly look at vision based processing AI. In this area, there are many companies that offer everything from chips to complete stacks specifically for this market. We will take a close look at where our strengths lie in the MCU and MPU space and focus our AI activities on that as well. Once that has been decided, we will help our customers port the AI models to our microcontrollers and microprocessors.

So the acquisition of Reality AI is an absolute plus for Renesas …

Absolutely. We continue to work with other ecosystem partners as before for other areas of AI model development. So this combination of combining AI models with our solutions is not the only one that we want to use or are already using. But in all cases, it's about completing our offering. Now with Reality AI on the model development side for the edge, we are meeting the trend that more and more intelligence is required on the edge. On the edge, the demands on computing power are not as high as in the cloud, but low power consumption plays a decisive role and thus the question of how the AI models can be optimized so that they fit the compute budget that is available.

Through the acquisition, we can implement the requirements just described more quickly. As mentioned earlier it doesn't mean we don't have other AI partners in our ecosystem. One example is Cyberon. We work with this partner on speech-based AI. Reality AI used to be a partner as well, but in this case we decided that it would be beneficial if we acquired the company. Because by acquiring Reality AI, we can build important competencies. There are few companies in the market that have the expertise for the industrial market like Reality AI. This was extremely important for us, so we are convinced that the acquisition is the right way to go.

You mentioned Cyberon and speech-based AI, this is about keyword recognition, not natural language processing, right?

Yes, natural language processing will continue to take place in the cloud, because the computing overhead is enormous. There are already systems that can recognize keywords, Cyberon goes one step further, because in that case it's about phrase recognition. I am convinced that this function will be needed more and more often at the endpoint, but if it is a matter of processing fluent spoken language, it will certainly take many years before this can take place at the edge, despite enormous progress in terms of the processing power of components that are optimized for applications at the edge.

Back to Reality AI, the company also maintains various partnerships with companies that compete directly with Renesas. How will that be handled?

That has not been finalized at the moment, but I expect that Reality AI will definitely be more focused on supporting our platforms than any others in the future as part of Renesas.

Reality AI has so far focused on being hardware agnostic, so that should change?

That’s the plan, although I would add that hardware agnostics in particular have been very attractive to us. But clearly, over time, we will integrate Reality AI’s models into our e2 studio as well as our translation libraries that we provide to our customers anyway. The goal is for it to make no difference at all for developers to use Reality AI expertise, even after we have acquired the company. This means that if a customer comes to Reality AI and says that they need to solve a problem with anomaly detection, for example, then this problem will be solved, but in the future our MCUs and MPUs will also play a decisive role in being implemented in the customer's system solution.

So you are convinced that Renesas gains decisive advantages by being able to offer Reality AI software that is optimized for Renesas components?

Absolutely. There are many industrial companies that want to solve problems with AI. But typically these are not AI or hardware specialists, these companies are about solving their specific problem. Just one example: a mining company. They want capabilities for screen wear prediction as soon as they need it, but they don't want to have to worry about PCBs, reference designs or anything like that, they just want to know how to figure out when to replace the screens. So they're looking for a company that can provide them with a complete solution for exactly their problem, where they're able to isolate faults and figure out when to change their screens, for example. In this instance they can use Reality AI Regression model to predict thickness of panel based on vibration during shaking of ore. Reality AI offers us decisive advantages exactly here, i.e. with customers who are not really interested in electronics, but focus exclusively on their core business. They're also not interested in the hardware that everything is built on, but only in getting their problem solved. And that's what we can address better with Reality AI, because this company works with end customers to solve their problems.

The same applies to other examples such as HVAC manufacturers. They want to know, if there is an increase in vibration in the system, which element of the system they need to replace before it fails. This is exactly the type of customer we can address directly with the acquisition.

That means with the acquisition of Reality AI, Renesas is also expanding its application know-how, right?

Exactly. But there is another important point: Reality AI also offers its tools in Japanese, and the company already has Japanese customers from the industrial segment. This fact also has definite advantages for us, because some major industrial and automotive customers are known to be based in Japan. Of course, there are also important customers in Europe who are also very pleased about this acquisition, but nevertheless the acceptance in Japan also brings advantages for a company like Renesas.

Anbieter zum Thema

zu Matchmaker+

  1. AI - step by step becoming a full-service provider
  2. A lot of activities in the AI area, but also in other areas, for example with Tata

Verwandte Artikel

Renesas Electronics Europe GmbH