ZF ramps up R&D spend on autonomous driving technologies

By Matthew Beecham | 14 February 2018

Torsten Gollewski

Torsten Gollewski

ZF recently presented its next steps on the road to autonomous driving. Its supercomputer, ProAI, acts as a central control unit within a test vehicle and with this ZF is taking a modular approach to the development of automated driving functions. The goal is a system architecture that can be applied to any vehicle and tailored according to the application, available hardware and desired level of automation. To find out more, we spoke to Torsten Gollewski, Head of Advanced Engineering at ZF and Managing Director of the ZF venture capital company, Zukunft Ventures GmbH.

Could you summarise the stage your ProAI has reached?

The ProAI Control Unit developed in partnership with NVIDIA is now production-ready and is already in development for a number of applications. This includes our partnership with Baidu where we have already demonstrated an automated valet parking system on a fleet of Pand Auto vehicles, one of the biggest Chinese car-sharing providers. In addition to the first generation ProAI we are already well into the development of a second generation unit with significantly higher processing power.

With its open architecture, ZF ProAI is highly scalable – the hardware components, connected sensor sets, and functional software can be adapted to the desired purpose and degree of automation from today's Level 2 and 3 automated functions and beyond to Level 4 and above applications. For example, ZF ProAI can be configured in terms of computer performance for almost any specific requirement.

In the application we showed at CES, the control unit uses the Xavier chip with 8-core CPU architecture, 7 billion transistors and the corresponding performance data management. It manages up to 30 trillion operations per second (TOPS) with a power consumption of only 30 watts. The chip complies with the strictest standards for automotive applications – just like ZF ProAI itself – creating the platform for artificial intelligence and deep learning.

By this time next year, what commercial success do you hope to have achieved with ProAI?

We are already making good progress on the commercial industrialisation of ProAI. We recently announced that an agreement has been reached with Chinese manufacturer Chery to deliver ProAI units to them by 2021. We are also working with our partner e.Go Mover in Germany where the first units of their automated electric commuter vehicle will begin testing and deployment this year.

We will also work with NVIDIA and DHL on a project that will utilise ProAI.  Duetsche Post utilises a fleet of 3,400 StreetScooter electric delivery vehicles, which can be equipped with ZF's multiple sensors including cameras, LiDAR, and radar that feed into the ZF ProAI system. This can enable the vehicle to use artificial intelligence to help it understand its environment, plan a safe path forward, proceed along a selected route and park itself – helping to ensure that deliveries can be made with greater accuracy and safety, and at lower cost.

Finally, we will start production of the Pro.AI next year for an off-highway equipment manufacturer.

ZF is best known for its chassis and transmission systems. But the company will always have to innovate or become a shadow of itself. Are you increasing your R&D spend in order to stay at the forefront of autonomous driving technologies?

We have increased R&D spend and expect that trend to continue.

Most certainly we have increased R&D spend and we expect that trend to continue – ZF invests around 6 percent of sales in R&D and engineering annually and we are shifting some of this spend from traditional products to the future megatrends of advanced safety, autonomous driving and electrification of the powertrain. We realise that these trends and technologies demand a high investment profile at this early stage – and we are also relying on our ecosystem of partners to help us achieve systems that will meet the mobility challenges of the future.

The systems that currently drive robot cars cost money. Lots of it.  It is said that the LiDAR is one of most expensive items. What's you view and how can the cost tumble?

LiDAR is one of the most interesting sensor technologies being developed to support automated driving.

LiDAR is one of the most interesting sensor technologies being developed to support automated driving, and to accelerate our time to market ZF recently took a 40 percent stake in the German LiDAR maker Ibeo, who have recently launched the industry's first mechanical scanning LiDAR system in production.

We are working with Ibeo on the next generation, solid-state scanning LiDAR - solid-state technology is less expensive than mechanical scanning LiDAR and also better from a packaging standpoint. And we are of course keeping a close eye on other entrants into the LiDAR market – some who claim they can decrease the cost significantly while also increasing the performance.

As with radar 20 years ago, as a new sensor technology evolves, costs can fall quickly as manufacturing volumes increase (enabling more efficient manufacturing processes), development expenses are spread across more volume, and silicon integration helps reduce component costs.

We understand that ZF will introduce Car eWallet in Germany in the third quarter of this year. How do you see this system being rolled out across Europe and to what extent are there still issues to be resolved?

Similar to the German roll out, the European rollout will be done in an ecosystem approach. Therefore cross-boundary effects of fleets that are operated in several European countries can be used.

Car eWallet will introduce the service domains in different releases thus being able to concentrate on each step. For example, we will be working with major ecosystem partners such as Europe's largest parking management company APCOA and Chargepoint, the world's largest vehicle charging network.  

What does the trend for shared mobility mean for ZF in terms of its products, technologies and manufacturing?

At CES, ZF unveiled a cloud-based platform for innovative mobility services.

The lesson for ZF and for everyone is to be flexible and open to new mobility models and new ways of doing business. In this space, we are also partnering to assist in the ways that urban transport and smart cities will shift the status quo. At CES, ZF unveiled its cloud-based platform for innovative mobility services. This cloud solution will make it possible to integrate different functions – ranging from fleet management and ride-sharing to innovative delivery services – into one shared intelligent platform, regardless of service provider.

When developing its initial application samples, ZF has closely collaborated with start-ups and innovative mobility suppliers. This Internet of Things (IoT) solution is being built on Microsoft's Azure platform - which offers a multitude of interesting and intelligent algorithms, but also the largest portfolio of security certificates industry-wide. This digital platform stands out due to its flexibility and scalability across industries. All business partners can decide which functions they want to use.

Is there still a lot of work to do to improve sensors for autonomous driving? 

ZF and its partners are relentlessly pursuing greater environmental sensing performance to ensure the most robust 360-degree view around the vehicle. We feel the best solutions are to use a combination of sensing technologies, using sensor fusion to compare the data among the sensors to help make the right decisions in terms of safety or autonomy.

Radar has long been a great complement to camera technology as it is virtually unaffected by snow or other weather conditions – and LiDAR can also be an excellent complement for accurate 3D vision of a scene – but can also be affected by weather, fog and certain light conditions.

We feel these sensors continue to improve and are reaching more optimum performance but there is always the possibility to improve. We can do this by fusing data from onboard environmental sensors together with localisation tools including GPS and high-definition maps.

How will the autonomous car change the look and feel of the cockpit?  

There is plenty of speculation about the cockpit of the future and how autonomous vehicles will transform personal and public transportation spaces. We believe it is safe to say that in the vast majority of cases the steering wheel and pedals are not going away completely anytime soon. As Level 3 systems that require handover from the vehicle to the driver and back again are employed, we will need smarter steering wheels and other human-machine interfaces that help the driver know when to take control and make the handover more intuitive. The ZF advanced steering wheel concept shown at CES uses gesture control, hands-on detection, an integrated and interactive centre screen and clever lighting to help make clear the current driving mode while presenting innovative ways to change vehicle controls like the HVAC or audio system through touch-sensitive technology.

Drivers often find it difficult to activate, set and monitor the multitude of assistance functions in today's vehicles.

The Cockpit 2020 concept we also presented is a new interaction concept which can help to relieve driver stress and prevent accidents. Drivers often find it difficult to activate, set and monitor the multitude of assistance functions in today's vehicles and ZF is addressing this challenge by showing the status of all assistance systems in one overview.

Furthermore, with this concept, ZF's latest integrated safety systems can prompt the driver to act when necessary. For example, we can use the Active Control Retractor motorized seatbelt to give haptic feedback to the driver through vibrating or tugging the seat belt – thus drivers can be warned to return to the driving task when their hands are not on the wheel. This also raises an important point: that seat belt and airbag technologies will be required for decades to come and will need to evolve as cockpits evolve and allow for other activities such as swivelling the seats for ease of conversation or allowing for more relaxed positions for resting, sleeping, etc. Here the trend is for these technologies to become increasingly integrated into the seat so that the protective technologies move with the seating position. ZF is working with interior experts partner Faurecia on such new interior safety solutions.  

More on AD, ADAS can be found in just-auto's research arm - QUBE