The autonomous driving & automated reality community gathered at the Automotive Tech.AD 2017 in Berlin to discuss the challenges of AI supported, cognitive cars in regards to complexity and penetration of in-vehicle infotainment systems and ADAS. The event was packed with inspiring keynotes and use cases – we put together the most exciting presentations and insights to provide a comprehensive review of Automotive Tech.AD Berlin.
Part 2 of the Automotive Tech.AD Berlin Diary introduces Bosch’s solutions for the development of automated traffic and shows new approaches and development processes for realistic traffic situations by sensor manufacturer Ibeo.
Bosch: A Scalable Architecture and Function Path from Assisted to Automated Driving
The two Bosch managers, Dr. Werner Uhler and Dr. Thorsten Huck, addressed the question of how scalability and functions lead from assisted to autonomous driving. Resulting challenges are manifold as vehicle safety is mandatory to be considered and different degrees of automation have to be distinguished. Driving assistance systems such as ACC are already available as a standard feature, but they can only check the longitudinal side.
Roadmap of Automation
According to the Function Roadmap, integrated driving assistance systems will be part of the equipment later this year. These include lane-keeping assistants that are functional up to 130 km/h and partly automate the parking process. From 2019 the motorway assistant will be available which is partly automated with lateral and longitudinal control functions. This level of automation allows lane changes after a confirmation and observation of the surrounding traffic and infrastructure, such as curves.
According to the two speakers autonomous driving on motorways will be possible around the year 2021. By then the environment and the complexity of the traffic can be reliably recognized, allowing the driver to be inattentive without causing any danger. Level 5 (full automotive) will be developed after 2025, says Bosch. Safety measures would have to be restrictive as there is no human control authority anymore.
Modular Systems and Re-Use
A study conducted by McKinsey shows a drastic shift in value creation processes. In the next 12-15 years hardware, software and engineering efforts will increase while the sectors software and engineering will grow much stronger than the hardware effort.
Re-Use is intended to increase the added value. Hardware can be reused, the software enables integration. The re-use concept may supported by OEMs, e.g. in context with ECU performance, security levels and communication bandwidth on the technical side. The Bosch experts suggest a combination of measures in order to avoid unnecessary costs.
The modular system means a possible hardware evolution, as for sensor technology. Level 2 automation already implies sensors to detect objects at the rear, side and front as well as traffic lights or more distant vehicles. Level 3 however provides greater precision and therefore needs more information about landscape features or free routes.
Combining Sensors
Sensor technology started to development radar systems for long distances forward. Nowadays radio detection is implemented for medium distances plus rear and side view. For level 3 automation stereo cameras and LiDAR technology are added to the list. Combining these sensors helps meeting accuracy and stability requirements. Same applies for driving assistance like ACC and lane-keeping assistants which are condensed to Integrated Cruise Assist.
Bosch integrates both the decentralized and the central architecture. Here again a combination of the two variations is used. The scalability makes it possible to offer in different packages – from basic to premium, whereby aging systems can be reused for a lower package.
Ibeo: New Approaches and Development Processes for Common and Realistic Traffic Situations
The path to successful autonomous driving is paved with good sensors, says Max Schmidt, Group leader for the development of Fusion Systems at Ibeo. Latter has started operating in the sensor business in 1998. In 2007 Ibeo participated in the DARPA Urban Challenge and managed to conclude corporations with Valeo, ZF Friedrichshafen and Audi over the years.
The Challenge of Localization: GPS vs Lidar
The problem about driving with GPS is that the positioning is too imprecise for autonomous driving. Max states that, if you only rely on GPS, the computer would quickly find itself on the wrong road. But with LiDAR you can create an HD card, which is used offline for position adjustment. This also makes highly autonomous driving possible.
Meanwhile the LiDAR system has now been further developed into a solid-state laser scanner. Thanks to a vertical image composition the system can provide a higher image resolution. By putting everything together - precise map data, an overview of the road, the tracks and the further route - the whole scenery can be interpreted.
Lidar finds use in complex traffic scenarios, such as encounters with pedestrians, because of its ability to collect the necessary card data, with the system also orientated towards relevant landscape features.
Ibeo’s Testing Environments
The data collected by the LiDAR sensor in the vehicle is processed (object recognition and classification), stored and sorted. Instead of human resources a reference sensor is used for validation purposes. The data is compared with each other in order to prevent possible misinterpretations. Ibeo is performing appropriate tests within simulations but also in the real environment. For tests in the environment Ibeo uses a fleet of 6 vehicles which are on the way in China, Japan, Australia, Poland and the US.
And that concludes part 2 of the Automotive Tech.AD Diary. We hope you enjoyed the contributions as much as we did - the next chapter will involve presentations by GM and Audi. If you would like to read the first part of the Automotive Tech.AD overview, you can access it here.