Autonomous vehicles (AVs) represent a major advance in transportation technology, promising to reshape the way people travel by providing a safer, more efficient and highly automated driving experience. However, as autonomous vehicles become more complex and operate without human intervention, ensuring their safety has become a key issue.
Microcontrollers (MCUS) are often at the heart of self-driving cars and are the brains of these systems. The MCU manages everything from data processing to controlling various sensors, actuators, and safety mechanisms.
As autonomous systems become more advanced, the reliability of MCUS is critical to maintaining operational safety. Even a small glitch in an MCU can lead to a system-wide failure that puts passengers and other road users at risk.
The lock-step architecture is used in the MCU to improve system reliability by detecting and correcting real-time errors. In a lockstep system, two or more of the same processing cores execute exact instructions simultaneously. The outputs of these cores are constantly compared to make sure they match.
Differences between cores indicate that an error has occurred, prompting the system to take corrective action. This redundancy ensures that minor failures are caught before they affect system performance, making it particularly valuable in safety-critical applications such as autonomous vehicles.
Traditional lock
Traditional lock-step systems rely on the same kernel executing the same instructions in parallel. The outputs are compared for consistency.
Using the same kernel to execute the same instructions makes the system vulnerable to common-mode failures. Two kernels may experience the same failure at the same time, resulting in ineffective error detection.
More advanced solutions have emerged to overcome the limitations of traditional lock-step designs.
A variety of lock step technologies
Unlike traditional lock-step architectures, which rely on the same kernel to detect errors, the diversified lock-step technique uses two different kernels that work differently. They use different architectures and instructions to accomplish the same overall task.
This diversity provides superior error detection capabilities and effectively reduces common mode failures. The variety of lock-step technologies is ideal for autonomous vehicles where safety and reliability are critical.
A variety of locking steps to improve security
A variety of lock-step technologies can be employed in the vehicle's electronic control unit (ECU), which manages real-time data processing and control in the autonomous driving system. As the two cores perform their tasks, the system constantly compares their output.
If the results of the two cores differ, an error has occurred. The system can then take corrective action, such as shutting down faulty processes or switching to backup mode to maintain security. Such continuous fault detection is crucial for autonomous vehicles, which rely on ECUs to process constant data from sensors, cameras, and other inputs.
Anomalies are detected in real time, allowing the vehicle to operate safely even in the event of minor hardware or software failures. The real-time nature of error detection also means that the system can respond immediately, preventing potential hazards.
Diverse cores improve fault tolerance, ensuring subtle or complex errors are identified and resolved before they affect vehicle performance. This reduces the risk of system failure, which is essential to maintaining the safety of passengers and other road users. By reducing common mode failures and improving overall reliability, diverse lock-step architectures enhance the robustness of autonomous vehicle systems, making them safer and more reliable.
Applications in self-driving cars
Multiple lock-step technologies are already being integrated into safety-critical systems in modern autonomous vehicles. Using different locking steps can reduce the development effort of ASIL-D systems. It is invaluable in ADAS because real-time data processing and fault detection are essential for safe operation.
By improving error detection and fault tolerance, the technology helps ensure that autonomous vehicles can safely navigate complex environments, such as high traffic or bad weather, making them more reliable in real-world scenarios.
Another key advantage is the ability of various lock-step processors to improve network security. Diversity in design and code minimizes potential vulnerabilities that hackers can exploit.
Future outlook
As technology evolves, continued research into lockstep systems may lead to more sophisticated error detection and fault tolerance mechanisms. With advances in processing power and safety architecture, future generations of autonomous vehicles will benefit from increased levels of safety and operational efficiency.