Ensuring Safety With Verification & Validation in ADAS

The automotive industry has seen significant strides in advanced driver assistance systems (ADAS). These systems, which include features like intelligent speed assist, adaptive cruise control, lane-keeping assistance, and automatic emergency braking, are designed to assist drivers and enhance vehicle safety.

While ADAS plays a crucial role in improving vehicle safety, ensuring the reliability of these systems is just as important as developing new features. That’s why verification and validation (V&V) processes are critical in ADAS development. They help engineers confirm that systems operate correctly under all conditions. Without rigorous testing, ADAS may fail in real-world situations, leading to safety risks and compliance issues.

Understanding Verification and Validation

Verification and validation are two distinct but interconnected processes in software and system development.

Verification involves assessing a product, service, or system to determine if it aligns with the established regulations, requirements, and specifications from the beginning of the development phase. This internal process focuses on examining documents, designs, code, and programs to confirm they‘re constructed correctly according to the defined criteria.

Since verification does not require executing the code, it’s often referred to as static testing. Common verification activities include document reviews, walkthroughs, inspections, and manual analysis.

Validation ensures that a software product fulfills the needs and expectations of its intended users within its intended environment. Unlike verification, validation is an external process that involves testing the actual product to confirm that it functions correctly and meets user requirements. Since validation requires executing the code, it is often known as dynamic testing. Common validation techniques include black box testing, white box testing, and various types of nonfunctional testing.

Challenges in ADAS Verification and Validation

Despite its importance, ADAS V&V presents several challenges due to the complexity of the technology, safety standards, and real-time performance requirements. Below are some of the key challenges engineers face during ADAS testing.

Complexity of Software and Algorithms

ADAS relies on a combination of sensors (radar, LiDAR, cameras), real-time data processing, machine learning models, and control algorithms. These components must work seamlessly to interpret the environment and assist drivers accurately. Since ADAS must function across different weather conditions, lighting levels, and traffic scenarios, engineers must simulate and test thousands of possibilities to ensure performance reliability.

With such a high level of complexity, undertaking manual verification of every possible system behavior is nearly impossible. So, a team of automotive OEMs, tier 1 suppliers, and research institutions got together and created standard ISO/PAS 21448, also known as SOTIF (safety of the intended functionality).

The standard addresses safety concerns in ADAS that go beyond traditional functional safety, ISO 26262. Unlike ISO 26262, which focuses on preventing failures due to hardware and software malfunctions, SOTIF deals with hazards arising from system limitations, sensor misinterpretations, and unforeseen operational scenarios, even when there are no system defects.

Engineers also need software testing tools to:

• Manage complexity.
• Confidently test the vast numbers of scenarios efficiently.
• Detect potential failures before vehicles hit the road.

Compliance With Automotive Standards and Regulations

ADAS compliance is based on various standards.

• SOTIF for key safety
• ISO 26262 for functional safety
• Automotive SPICE for software development maturity

ISO 26262 mandates rigorous V&V processes along with risk-based automotive safety integrity levels (ASILs). Nevertheless, with advancements in automation and the growing use of automated driving systems (ADS) in self-driving technology, new regulations continue to emerge.

For example, UN R171, effective since September 2024, governs Driver Control Assistance Systems (DCAS) in 59 countries, emphasizing driver engagement and system performance. It establishes clear requirements for driver monitoring, system interventions, and operational safety.

With all these standards and regulations, satisfying compliance is a complex and time-consuming process. It requires rigorous documentation and extensive testing. Engineers can streamline this effort by integrating continuous testing into their CI/CD pipeline, leveraging test automation solutions like the following to help satisfy compliance requirements more efficiently:

• Static analysis
• Unit testing
• Structural code coverage
• Requirements traceability

Each standard has distinct requirements but overlaps in key areas such as safety analysis, software development maturity, and risk mitigation. Detailed documentation, like safety plans, technical safety requirements, and software safety cases, is critical with traceability from requirements to test results.

The following are essential for ensuring compliance with these standards.

• Maintaining detailed documentation and traceability across all stages of development.
• Following a structured V-model process with an Agile development methodology.
• Performing regular safety audits and independent assessments.

Real-Time Performance in Diverse Scenarios

ADAS systems operate in real-time and process massive streams of sensor data to make split-second decisions. A delay in detecting a vehicle, pedestrian, or road hazard could lead to accidents. Validation must, therefore, test ADAS in a wide range of conditions, such as:

• City streets with heavy traffic
• Highways with varying speeds and merging vehicles
• Sudden pedestrian crossings
• Extreme weather conditions, such as fog, heavy rain, and snow

Building a robust testing framework capable of replicating real-world driving scenarios accurately is challenging. ADAS engineers rely on simulation-based validation and hardware-in-the-loop (HIL) testing to create real-world test environments without requiring millions of on-road miles.

Parasoft C/C++test supports HIL testing or what we also internally refer to as cross-platform testing. While HIL testing can be part of integration testing, not all integration testing involves HIL.

Software integration testing can occur purely in a virtualized environment, while HIL testing specifically involves real hardware interactions with simulated environments. This offering and capability helps with shift-left target testing, aids in increasing code coverage, and can be integrated into CI/CD pipelines.

In addition, collecting test results and code coverage data from the target system is essential for validation and standards compliance. Target hardware may have limited physical connectivity making it more difficult to retrieve test results. Fortunately, Parasoft’s C and C++ testing solutions make it possible to extract the data from various ports, such as serial, Ethernet (TCP/IP Sockets), JTAG connector, and other methods.

Challenges With AI and Machine Learning in ADAS

Many ADAS features, such as adaptive cruise control, lane departure warnings, and object detection, use AI/ML to improve decision-making. AI’s ability to process vast amounts of sensor data in real time makes it indispensable for modern vehicles.

Autonomous driving systems, for instance, rely on AI to interpret inputs from LiDAR, cameras, and radar to navigate complex environments. Beyond autonomy, AI enhances predictive maintenance, personalizes in-car experiences, and optimizes energy efficiency in electric vehicles. While AI enhances vehicle perception and automation, it also introduces new V&V challenges in satisfying safety, security, and certification requirements.

However, companies like Waymo have demonstrated the use of AI and their achievements in overcoming V&V challenges with AI nondeterminism safety concerns, security vulnerabilities, real-time constraints, and certification.

Today, Waymo, with their self-driving fleets, are expanding globally. AI-driven features like adaptive cruise control and lane-keeping assist are now mainstream, underscoring consumer demand for smarter, safer vehicles.

To help address these challenges, engineers need to understand what is artificial intelligence in software testing and how to satisfy strict safety and security compliance demands when using AI and ML in ADAS.

How Parasoft’s Testing Solutions Ensure ADAS Verification and Validation

Parasoft’s testing solutions ensure ADAS verification and validation by:

• Automating critical testing processes.
• Improving software quality.
• Ensuring compliance with industry standards like ISO 26262, ISO 21448 (SOTIF), and automotive SPICE.

ADAS software is highly complex, requiring rigorous testing to handle real-world driving scenarios, edge cases, and safety-critical functions.

Parasoft’s static analysis helps enforce coding standards such as MISRA, AUTOSAR C++14, and CERT C, identifying potential defects early in the development cycle. Static analysis reduces the risk of software failures that could compromise ADAS performance by detecting issues related to memory management, concurrency, and undefined behavior.

Unit testing and structural code coverage further enhance V&V by ensuring that individual components function correctly and meet safety requirements. Parasoft supports automated test generation, fault injection testing, and mutation testing, improving software robustness.

For integration and system-level validation, Parasoft provides requirements traceability and compliance reporting, ensuring full visibility from requirements to test execution. This is essential for demonstrating compliance with ASIL requirements under ISO 26262. Additionally, HIL integration enables verification of ADAS algorithms in simulated environments before deployment in real-world conditions.

Parasoft’s solutions also support test automation in CI/CD pipelines, enabling continuous validation of ADAS software throughout development. This integration ensures rapid detection of regressions and helps teams maintain high software quality while accelerating product releases.

ADAS developers improve safety, reduce verification effort, and accelerate time to market for advanced driver assistance systems with comprehensive automated testing solutions, compliance verification, and risk-based testing strategies.

Conclusion

Verification and validation are critical in ADAS to maintain safety, standardization, and public trust. However, traditional testing approaches struggle with the increasing complexity of V&V in ADAS.

ADAS teams can simplify and strengthen verification and validation in ADAS with automated testing solutions like Parasoft’s that detect defects early, improve compliance, and enhance overall system performance.