This week Arm is visiting the IAA Motorshow in Frankfurt, Germany. We exhibiting in the New Mobility World Conference, showing how Arm and our partners technologies are leading the digital transformation of the car. For those of you who can't make it to the show, here is a roundup of the great demos from our partners.
To demonstrate the future of in-car experiences, we are exhibiting a market-ready, scalable infotainment and cockpit control system from MediaTek and Mobica, featuring TomTom Navigation.
This eCockpit system is demonstrating a Cockpit Domain Controller, which brings together multiple in-cabin functions to run on one single System-on-a-Chip (SoC). Until now, different cockpit systems, such as the instrument cluster, the In-vehicle Infotainment (IVI) head unit and park assist Surround View Monitor (SVM) systems have required individual SoCs to implement their functionality, which means extra cost and development complexity.
This demonstration shows the instrument cluster, IVI and SVM systems running on a single, high-performance SoC: the MediaTek Autus I20 (MT2712). It is a highly capable hexa-core CPU featuring four high efficiency Arm Cortex-A35 processors and two high performance Cortex-A72 processors together. This CPU complex is paired with a high-performance Arm Mali-T880 MP4 GPU. The Autus I20 can support up to three displays for different uses around the vehicle, which in this case is the instrument cluster, the IVI and the SVM.
Mobica have developed the cluster HMI and infotainment software solutions for this hardware platform. The Mobica cluster HMI, running on Linux, shows the vehicle speed, with responsive rev counter and other driver information, such as warning indicators and turn signals, and includes safety mechanisms to enable an automated recovery in case of an error.
The Mobica infotainment software solution runs on Android, providing the user with functionality such as media playback, vehicle settings and phone connectivity. The infotainment display features TomTom connected Navigation, with intuitive destination entry, moving lane guidance, interactive 3D map visualization and TomTom’s accurate Traffic service
The SVM system is provided by MediaTek SVM. This system seamlessly knits together views from multiple cameras to give the driver a clear picture of the car’s surrounding and any obstacles or hazards close to it, aiding the driver during maneuvers and parking.
This demonstration shows the rich functionality we can expect from future eCockpit solutions and the packaging and simplicity benefits from having all three systems delivered by a single, reliable SoC.
Also on the stand at IAA is a demonstration of Starling from Swift Navigation. Earlier this week, we announced were partnering with Swift Navigation to bring a cost-effective, scalable and high-integrity positioning solution to developers of autonomous and connected vehicles.
As autonomous vehicles become more of a reality, it will be critical to the safety of the passenger for a self-driving car to be able to accurately determine its position in the world. Navigation systems are commonplace on the cars of today, but these types of systems are only accurate to around 3-5 meters, which is not sufficient for lane-level positioning accuracy. Accuracy is not the only requirement here – high-integrity measurements are critical to meeting the strict safety requirements of autonomous vehicles, demanding full confidence in that positioning.
The Starling positioning engine is an Arm-based GNSS solution. This works in conjunction with a correction network of base stations to enhance the measurement accuracy of commercially available GNSS hardware solutions. This helps Swift enhance precision down to the centimeter. Starling is both a hardware proven and hardware agnostic, end-to-end solution, tunable for customer specific requirements and designed to be compatible with industry leading silicon makers who build their solutions on Arm. As well as its lane-level accuracy, it is also a high integrity solution with end-to-end redundancy and integrity monitoring services.
The demo we are showing at IAA is using some pre captured data positioning data, from driving around the roads near to Arm headquarters in Cambdridge, UK. It is running on an Telemaco SoC from ST featuring an Arm Cortex-A7. The demo shows the position of the car updated on a map to lane level precision, which you can verify by watching the video captured at the same time.
Finally, we have Brodmann17 showing their perception technology for autonomous driving. Their goal is providing the most efficient perception solution possible for different types of ADAS product without compromising on quality and accuracy.
Brodmann17’s patented technology provides the next generation of perception for automated driving solutions. Their revolutionary Deep Neural Network (DNN) architecture requires only a fraction of the usual computing power and can run highly- accurate computer vision on low-power Arm processors, ideal for mass production. Their software architecture achieves state-of-the-art accuracy but consumes only a fraction of the computing power, making it accessible to the mass market, providing a more efficient and cost-effective deep learning perception solution with 20x efficiency compared to competing ML algorithms.
In the demo at IAA we have here is Brodmann17’s Forward Collision Warning (FCW) application running on an Arm applications processor, utilizing only a single core while maintaining high performance and excellent accuracy.
Arm’s Automotive Enhanced CPUs, part of the Arm Safety Ready portfolio, are an excellent baseline for the Brodmann17 software and this synergy provides scalable robust perception.