Editor’s note: Um die deutsche Version dieses Artikels zu lesen klicken Sie hier.
Autonomous driving is coming. However since the big hype generated by prototypes of self-driving cars during the CES 2015 in Las Vegas we haven’t really seen any satisfying cars which don’t need the driver’s attention while cruising along. Sure Google’s cute little vehicles drive around in the Silicon Valley but they didn’t stay accident free even with top speeds limited to only 40 km/h. Tesla introduced an auto-pilot feature with the software update 7.0 for the Model S which is basically an upgrade of the intelligent Cruise Control and enables the car to steer itself. Impressive, but only useful on highways, speeds of up to 110km/h and with both hands in close proximity to the steering wheel in case you need to intervene.
The real dream of autonomous driving however looks different. We want to travel relaxed and rapidly without planning 12 hours to bridge the distance from Munich to Berlin (approx. 580km). We also don’t want to get a heart attack every time we have to step in when our car does something stupid.
What would be necessary to make the vision of autonomous driving reality? Cars need to be able to see and react better. To achieve this, intelligent sensors and camera systems which can observe the complete surroundings of a car are required.
The Myestro Interactive GmbH is working on a solution that is capable to tackle this challenge. The Karlsruhe, Germany based technology company develops products for diverse markets from consumer to automotive customers. Myestro’s solutions are based on the 6D-Stereo technology which is at the core of innovations like the industrial camera system mvBlueSIRIUS from Matrix Vision, which is also capable of calibrating itself thanks to RubberStereo technology.
Myestro’s sophisticated AirCrown gesture control is able to recognise the user’s hand even if he is holding something in it or if it is behind a shop window. This solution is already deployed in market-ready applications.
Myestro has conducted a lot of research within the field of self-driving cars for example during the Prometheus Project. The extensive experience of more than 30 years in this field in combination with 6D-Stereo and RubberStereo enable the patented interlinking of stereo cameras for the surround view of 0-50 meters.
Together with EBV’s support Myestro is now developing a camera based environmental sensor system for autonomous vehicles called “Safety Belt”. We talked with Dr.-Ing. Jens Schick, CTO with Myestro, and Frank-Steffen Russ, Director Vertical Segment Automotive at EBV in order to get some deeper insights on this amazing project.
Can you describe Safety Belt in one sentence?
Frank-Steffen Russ: Safety Belt describes the approach to link intelligent autonomous cameras to one environmental sensor system for the application in the fully autonomous vehicle.
What is the goal you want to reach with Safety Belt?
Jens Schick: Undoubtably autonomous systems need sensors to observe their environment. Camera systems play a major role within the equipment of self-driving vehicles together with radar, lidar and car-to-environment-communication. It is predictable that we will see autonomous (robot) vehicles going into production soon. If we use camera systems to achieve this, then we will have to answer the question to what extent they are functional safe and redundant. We will have to figure out which approach is best in order to add functional safety to camera based environmental sensor systems. In addition the integration of cognitive intelligence will be an important task. The Safety Belt project aims to not only answer those questions but to deliver solutions.
Ok – it is our understanding that functional safe advanced driver assistance systems (ADAS) already exist. What is so special about Safety Belt in this context?
Jens Schick: Functional safety is closely aligned to the requirements and status-quo of technology. At Myestro and our partners we obtain the know-how which is necessary to access the next level of environmental sensor technology. Today the intelligence is typically integrated in the ADAS-main computer or the sensor fusion unit. With Safety Belt we will pursue a different approach as the image sensors deliver already processed environmental data. This can only be achieved if the sensors are linked to a belt around the vehicle. The linkage to one redundant system is enabled by the RubberStereo technology which makes it possible to recognise if the position of the cameras in relation to each other changes and to react to such changes. The belt of environmental sensors takes a lot of processing tasks off the ADAS main computer as the required data is delivered in real-time and Automotive Safety Integrity Level D (ASIL-D) compliant.
How will the Safety Belt be applied in practice?
Frank-Steffen Russ: This technology is implementable into every vehicle or machine which requires environmental analysis and observation. For applications in the area of mobility, autonomous driving and in other machines this approach means increased safety and scalability. Here’s an example: In the area of commercial and agricultural vehicles (CAV) there are many vehicles which consist out of tractor and trailer. In this case the Safety Belt approach can optimise the environmental observation significantly as tractor and trailer collect data independently of each other. Tractor and trailer deliver this information either to their central units or can be linked to compile one unit.
How is EBV helping to realise this project?
Frank-Steffen Russ: The Safety Belt project will investigate different applications and equip them with state-of-the art hardware. That’s where EBV as semiconductor specialist steps in and delivers the perfect components along with the required design expertise. In addition we will make Safety Belt available to a huge customer base using our extensive third party partner network.
What are the next steps within this project?
Jens Schick: The project itself is already structured and defined. Currently we conduct a lot of research in order to determine which applications are the best fit for the initial development. In addition to cars and CAVs also transport machines of all kinds could be a possibility. As the common definition of environmental sensor technology and the Safety Belt approach differ it is necessary to work very closely with potential users and customers out of different areas. Today we are leading discussions with industrial partners in order to define the best use case.