Safety of Highly Automated and Autonomous Systems

Autonomous driving is one of the key future topics with the potential to bring disruptive changes to our mobility system. Social acceptance depends entirely on ensuring both technical and legal safety.

At our Mobility Innovation Campus (MIC), we contribute to the testing, development, and ultimately the assurance of the safety of mobility concepts involving autonomous vehicles. The trials planned here focus on the interaction with so-called Vulnerable Road Users (VRU), such as cyclists and pedestrians, as well as testing contactless (inductive) charging of electric vehicles.

We support and guide our customers in safeguarding AI applications and fulfilling the relevant requirements for approvability. This includes everything from the selection and evaluation of training data to the generation of application-specific test or simulation data, and the detailed performance analysis of the AI system.

Our experts are actively involved in standardization committees related to the certification of AI algorithms and machine learning (DIN Standardization Roadmap AI, ISO / IEC JTC 1/SC 42, ISO AWI TR 5469). For highly automated vehicles, we focus on safe and reliable object recognition using relevant sensor technologies.

• Evaluation of AI algorithms in terms of performance, robustness, explainability, and uncertainty (Uncertainty Quantification and Safety Cage for Computer Vision)
• Generation and preparation of suitable test datasets
• Process support and consulting for the validation of machine learning
• Dashboard with qualitative metrics and predicted safety level

The networking of road users with each other and with intelligent infrastructure plays a crucial role in the efficient organization of traffic and the improvement of road safety. Reduced latencies, made possible by 5G mobile communication technology, particularly enable safe interaction and cooperation between automated vehicles. In the context of connected driving, we support our customers in the design and safeguarding of communication technologies, as well as in the evaluation of robust network topologies.

• Analysis of the security level of connected driving with V2X communication
• Implementation of secure communication layers
• Securing communication channels with respect to the following protection objectives: integrity, authenticity, availability, and confidentiality
• IT security and investigation of vulnerability to disruptions
• Measures to maintain IT security during operation

As a BSI-certified IT security service provider, we support our customers in the prevention and protection against cyber attacks. The networking and automation of vehicles in autonomous driving increases the risk of on-board cybersecurity breaches. Wherever software with internet or mobile access is installed, there is a risk of hackers gaining unauthorized access to the system. Our cybersecurity experts assist you in the challenge of protecting your IT/OT infrastructures from such attacks.

• Implementation of ISO 27001, IEC 62443 and BSI IT baseline protection standards
• Conducting gap and threat analyses, as well as risk assessments
• Designing security concepts and secure network architectures
• Securing communication interfaces
• Secure integration of next-generation mobile networks and Galileo PRS
• Protecting security-critical functions from external interfaces

Despite advances in automation, people still have to take responsibility (see Human System Integration). Display and operating concepts must be developed in such a way that they distract as little as possible from the (assistance-supported) driving task. At SAE level 3, functions that contribute to driving safety at the human-vehicle interface play a decisive role. One problem that has not yet been solved is the driver taking back control of the vehicle at higher SAE levels in the event of a sudden failure of the automatic control system.

• Design analyses of human-machine interfaces (HMI) for assistance systems
• Investigation of usability (RAMSIS) and user experience
• Evaluation of controllability and foreseeable misuse
• Stress tests for takeover scenarios (takeover request)
• Interior monitoring and fatigue detection using eye tracking
• Advice on situational and mode awareness

An introduction to the topic Resilience Testing of Autonomous Systems / Jamming & Spoofing (Video) is provided by our consultant Maik Holzhey.

Global navigation satellite systems (GNSS) play a crucial role in automated driving, providing highly accurate positioning and time synchronization for the overall system. However, incorporating GNSS also introduces a potential threat from targeted cyberattacks on the satellite-emitted navigation signals.

• Integrity assessment of positioning using GNSS
• GNSS test laboratory for investigating jamming and spoofing
• Test infrastructure in Ottobrunn (indoor and outdoor)
• Development support for SLAM (Simultaneous Localization and Mapping)
• Sensor data fusion (e.g. GNSS, IMU, odometry, LIDAR, cameras, etc.)
• Mapping of position onto georeferenced HD maps

Simulation is essential for the development and validation of highly automated vehicles. By combining virtual test drives with real-world tests, testing efforts can be kept within economical limits. Scenario-based simulation offers significant advantages, especially in terms of reproducibility and scalability. However, the validity of the simulation in the applied use cases is crucial. We develop customer-specific simulation environments for the virtual testing of automated systems.

For the development-related testing of components and automated systems, we support our customers with tailor-made XIL test solutions (SIL, HIL, DIL, VIL). We integrate the desired simulation environments (including commercial solutions like IPG CarMaker, dSpace ASM, VIRES Virtual Test Drive VTD) into the testing environment, customized to meet specific customer needs.

• Development of a distributed simulation with integration of customer-specific models
• Modeling of the integrated complete vehicle and embedded technologies
• Validation of the simulation using available test data
• Digital Twin in the context of route approval
• Virtual environment including weather conditions using a game engine (Unity 3D)
• Georeferenced HD maps by IABG Geodata Factory

A prerequisite for automated mobility is the detection of the surroundings using reliably functioning sensor technology. Only through the combination of different sensor technologies such as cameras, radar, lidar and ultrasound, can critical situations be handled in adverse weather conditions. We support our customers in securing their sensor systems through sensor performance tests and provide services as part of product qualification. We operate test laboratories for conducting mechanical and climatic environmental simulations as well as EMC testing.

• Object detection under mechanical and climatic environmental influences
• Mechanical tests (vibration, NVH, shock)
• Modal analyses to identify natural frequencies (resonance)
• Climatic tests (temperature, corrosion, UV)
• Electromagnetic compatibility (EMC) testing