How Data Will Drive the Future of Intelligent Vehicles – Germany’s COMPREDICT CEO Stéphane Foulard

German start-up COMPREDICT develops virtual sensors that enable analysis of vehicle data at scale. Since its launch in 2016, the Darmstadt-based company has continuously grown. It now employs 25 domain experts, data scientists, programmers and business developers.

Auto Futures has been talking to COMPREDICT’s CEO, Stéphane Foulard.

Together with Rafael Fietzek, Foulet founded COMPREDICT after the completion of their PhD studies at TU Darmstadt. During research work, they recognised the potential of in-vehicle data, and what the intelligent use of it can generate in terms of insights.

“Even today, in-vehicle data analytics remain at an early stage in our industry. Large amounts of data are generated, but often without a precise methodology allowing efficient data analytics at scale. There is a demand for clear, powerful, and privacy-compliant tools that can be used across departments and easily rolled out,” says Foulet.

“Efficient use of in-vehicle data allows cost savings, emission reductions and additional revenues for manufacturers and fleet operators, which in turn transforms into benefits like less consumption or less maintenance costs for the consumers. This can be enabled through targeted component design, right-sized durability, predictive maintenance and optimized fleet management.”

Instead of hardware, the company’s virtual sensors are software-based.

“They use standardly available data in connected vehicles and process them with data-driven approaches like machine learning, to compute new insights which are otherwise costly, inconvenient, or impossible to measure,” he explains.

Examples of what can be measures are wheel forces, brake pad wear and battery anomalies.

Foulard went on to discuss the main advantages of virtual sensors over physical sensors.

“Firstly, laboratory-type hardware sensors used during vehicle development are precise; but they are costly, error-prone and require specific maintenance. Though they are necessary, they are not meant to be used at larger scale. We can train virtual sensors on one single prototype vehicle equipped with lab sensors, and then deploy software-based virtual sensors on a larger fleet of non-equipped prototype vehicles, allowing cost savings >90% for a fleet of 10 vehicles.”

“Secondly, virtual sensors can be used in standard series vehicles you see on the road. They can replace existing hardware sensors, thus reducing the amount of on-board electronics and costs, and they can enhance existing sensor data as they provide new insights like continuous vehicle mass, brake and tire wear or EV battery aging,” adds Foulet.

COMPREDICT’s virtual sensors are managed at scale in the cloud by its algorithm-hub called AI CORE. This company says this technology has the potential to help automakers unlock new and  future revenue streams.

“Vehicle manufacturers (OEMs) benefit from precise component load data that allow more efficient test campaigns and faster product development. Connected vehicle data analytics also paves the way for new usage-based revenue models in the domains of insurance, maintenance or road safety.”

“Fleet operators can leverage vehicle usage data for predictive maintenance and scheduled spare parts ordering. On top of this, improved fleet availability increases revenue and contributes to drivers’ satisfaction at work,” he says.

Entering The BlackBerry IVY Ecosystem

In June, 2022, COMPREDICT closed a €5.5 Million Series A funding round led by Vektor Partners and BlackBerry. It will use the funding to deepen its collaboration with automakers and fleet providers, as well as expanding into new markets.

“With our new strategic investor BlackBerry, we will now go into the direction of edge application, meaning implementing and running our virtual sensors directly in the vehicles. This way, we will drastically reduce the amount of data that must be sent to the cloud. This represents an undeniable enabler to scale our solution over millions of vehicles at lower costs,” says Foulard.

In a press statement, Anooj Shah, Senior Director, Investments & Strategy at BlackBerry, comments: “COMPREDICT’s industry-leading virtual sensing technology combined with deep automotive domain expertise matches BlackBerry IVY’s vision of enabling advanced vehicle data insights for automakers and the broader mobility ecosystem. We are thrilled to bring them into the BlackBerry IVY ecosystem to support their mission of making accurate predictive maintenance and vehicle diagnostics more accessible and enable automakers to transform these insights into new service offerings.”

“We are currently growing our technical team of domain experts, data scientists and software developers. This will enable us to address large scale fleets with vehicle OEMs, and to further automatize the workflow on the product side. In parallel, we will grow our teams in business development, marketing, and sales in order to further develop our business in Europe, East Asia and USA,” adds Foulard.

How COMPREDICT Can Lighten The Load For OEMs

France’s Michelin was an early investor in COMPREDICT. Its turnkey electric mobility solution, Watèa, integrates COMPREDICT’s virtual sensors into its electric mobility offer.

“Based on COMPREDICT’s mass estimator, Watèa by Michelin provides fleet managers with key indicators that allow them to fully exploit the potential of their electric vehicles. The purely software-based mass estimator enables fleet managers to select the right vehicle for each driver, estimate payloads and to optimize maintenance as well as charging.”

As well as cost savings, the company’s offering can also help reduce the environmental footprint of mobility over the entire product life cycle.

“Today, vehicle components are often over-designed ‘on the safe side’ so that they do not break over a whole vehicle lifetime, even for the most demanding drivers. But this does not make sense for over 99% of the drivers: over-design increases cost, weight, fuel consumption and CO₂ emissions,” says Foulard.

“We have demonstrated, how our data-driven design approach can enable a right-sizing, meaning a dimensioning of the vehicles according to the usage without increasing failure probability or generating on-cost for the consumers. For mechanical systems in a vehicle like a gearbox, this might represent weight reduction of >15%, the cost reduction of >5%, and reduction of CO2 emissions during manufacturing by >15%.”

Finally, we asked Foulard what transportation and mobility will look like by the end of this decade.

“Data-driven decision making and taking will be part of standard business for all stakeholders. Data will be an asset among others and, more important, the crucial part will be what you will generate out of the data. In any case, there will be a consolidation of the entire ecosystem comprising data exchange standards, data sharing platforms and business models.”

“End users, NGOs and industry will benefit from transparency throughout the value chain over the whole lifecycle of vehicles. Virtual sensing technology will play a key role in generating insights to reduce the environmental footprint in development, production, usage and repurposing of vehicles,” he concludes.