Researchers from the Politecnico di Milano, led by Pietro Dardano, Paolo Rocco, and David Frisini, have pioneered a novel approach to automate functional test procedures for both civilian and military aircraft certification. Their project, ARTO, combines collaborative robots, or cobots, with advanced Artificial Intelligence techniques to revolutionize avionics testing.
ARTO is designed to interact with cockpit components, recording forces, torques, and the positions of the cobot’s end effectors during these interactions. The raw data collected is first preprocessed to filter out disturbances caused by low-performance force controllers and embedded Force Torque Sensors (FTS). This preprocessing step ensures that the data fed into the subsequent machine learning models is clean and reliable.
The heart of the ARTO system lies in its use of Convolutional Neural Networks (CNNs) to classify the cobot’s actions as either Success or Fail. But the innovation doesn’t stop at classification. The CNNs are also trained to identify and report the causes of failure, providing valuable insights into the testing process. This capability is crucial for diagnosing and rectifying errors, thereby enhancing the overall reliability of the automated testing system.
To further improve the interpretability of the model’s decisions, the researchers have integrated Grad CAM, an eXplainable AI (XAI) technique. Grad CAM provides visual explanations for the model’s predictions, making it easier for human operators to understand and trust the automated system. This transparency is particularly important in the high-stakes environment of aircraft certification, where understanding the reasoning behind a decision can be as critical as the decision itself.
The practical applications of this research for the defence and security sector are manifold. Automated avionics testing can significantly speed up the certification process for military aircraft, ensuring that they are ready for deployment more quickly and efficiently. Moreover, the ability to diagnose and rectify errors swiftly can enhance the safety and reliability of military avionics, which is paramount in high-stakes operational environments.
In conclusion, the ARTO project represents a significant leap forward in the automation of avionics testing. By combining cobots with advanced AI techniques, the researchers have created a system that is not only efficient and reliable but also transparent and trustworthy. This innovative approach has the potential to revolutionize aircraft certification processes, benefiting both civilian and military aviation.
This article is based on research available at arXiv.