But what’s happening around the vehicle? One requirement for automated driving involves reliable and precise perception of the surroundings. As the vehicle is required to move autonomously in real traffic situations, it must be able to detect and localize all relevant road users within the entire vehicle environment (360 degrees). This entails several sensors applying various surveying principles to monitor each area of the vehicle’s surroundings. This improves the reliability and stability of information. Most of the sensors required for these tasks are already in volume production today. For example, partially automated systems are already available which support drivers in difficult situations. These include remote park assist, garage park assist, home zone park function, traffic jam assist or highway assist.
In specific cases radar, video and ultrasonic sensors are not sufficient to deliver plausible data of two sensors constantly. Which is why Bosch is working on new sensor technologies to meet the high demands on surroundings perception. The data provided by individual sensors are combined and processed to form a comprehensive environmental model representing all static and dynamic objects. Entirely new hardware and software technologies and new algorithms are used for this calculation.
A basic requirement for highly and fully automated driving is represented by vehicle localization which must be both exact and constantly available. A single sensor is unable to fulfill these requirements which is why a combination of surround sensors (for detecting landmarks such as lane markings and buildings) with satellite navigation and correction service as well as inertial sensors and a digital map is used. High-definition digital maps are used for this purpose which contain far more layers of information than the maps for standard sat-nav devices. They also support the automated vehicle in planning individual driving maneuvers, e.g. whether the vehicle should change lanes or not. Through connectivity with the cloud, the data contained in the maps are constantly updated enabling inclusion of dynamic factors such as temporary closed lanes in travel plans. For the so-called road signature, Bosch has developed a pioneering solution for creating key parts of high-definition maps using vehicles’ radar and video sensors, literally as they drive by.
Information on current road conditions is also required by automated vehicles in order to detect potential dangers at an early stage and to ensure safety during driving. For these innovative road condition services, Bosch uses weather data supplied by its partner Foreca in a first expansion stage, in order to draw conclusions about possible hazards such as aquaplaning, ice or snow. Automated cars can then adapt their driving behavior to the respective conditions, choose a different route or ask the driver to take control if a safe onward journey cannot be guaranteed in automated mode.
In the future, Bosch will be supplementing its predictive road condition services to include data from the vehicle. What inside and outside temperatures are measured by the car? Are the wipers activated? Connectivity means that such information does not stay unused in the vehicle but reaches the cloud through the respective manufacturer’s backend. Bosch also evaluates the control interventions by the ESP® skid protection system in order to determine the friction value and condition of the road surface. All data combined and evaluated intelligently result in a smart Bosch service – and a feeling of safe automated driving.
Highly and fully automated functions need to monitor the driver as well as the vehicle’s surroundings. After all, drivers are not obliged to monitor the system when these functions are in operation. They can pass full control to the system, at least for a certain period of time or for a defined situation, after which the driver is requested to take control of the driving task – and the vehicle must be able to detect if the driver is capable of doing so. Bosch has already developed systems such as driver drowsiness detection which constantly monitors the driver and issue warnings in dangerous situations such as fatigue or microsleep.
The system passing driving responsibility to the driver remains a challenge for engineers. How is the driver informed, how long does the driver need to take over? What happens if the driver does not take control? One possible scenario: If the driver fails to take over, even after multiple alerts, as the car approaches a freeway exit, the automated vehicle would stop autonomously on the hard shoulder.