A*STAR researchers are working on a new project called “AI for Airline Operations” under a joint lab between Singapore Airlines (SIA), SIA Engineering Company (SIAEC) and A*STAR, to develop advanced AI solutions that improve engineering productivity, customer experience and cost-effectiveness of airline operations through predictive analytics and optimisation.
A collaboration between SIA, SIAEC and A*STAR’s Institute for Infocomm Research (I2R) and Institute of High Performance Computing (IHPC), this project broadens the scope of an earlier project that leveraged AI for predictive maintenance of aircraft components.
Powered by data harnessed from state-of-the-art deep learning techniques, the innovative predictive analytics technology reduces the risks of flight delays by detecting recurring defects and predicting component failures in aircrafts. It also enhances SIA’s customer experience with personalised shopping recommendations generated from customer behaviour analytics, and detects anomalies in their loyalty programme transactions through fraud analytics. In addition, text analysis applied to manuals and guidelines for information extraction and building comprehensive knowledge bases allows for better Q&A systems and responses to customers’ queries.
Additionally, A*STAR’s optimisation technology helps SIA and SIAEC improve their operational efficiency. This is made possible through the optimisation of maintenance interval, workflow sequence, manpower and resource allocation. This three-year project is currently in the development phase. A*STAR’s AI expertise, combined with the technical and business expertise of the various SIA and SIAEC departments, are key to the development and delivery of the AI solutions. The parties also plan to progressively rope in local SMEs in the developmental process.
A*STAR researchers are developing enhanced AI and robotic solutions to improve the accuracy and speed of defect detection during pre-flight checks and airplane maintenance processes on airframe surfaces like the wings and undercarriage.
The enhanced technology reduces the defect detection time and the risk of defects being overlooked during manual human inspections with a more comprehensive scan coverage. Early detection also allows aircraft maintenance, repair and overhaul (MRO) companies to plan aircraft servicing in a more efficient and timely manner.
The Smart Automated Aircraft Visual Inspection System (SAAVIS) programme by A*STAR is at the centre of these novel solutions being developed. SAAVIS combines expertise in robotics, computer vision and AI from A*STAR’s Institute for Infocomm Research (I2R) with path planning capabilities from A*STAR’s Institute of High Performance Computing (IHPC).
The aircraft’s location and state are first captured by high-resolution cameras installed in the hangar and autonomous ground robots using 3D localisation technology, where the image data is processed using AI technology to automatically detect defects before notifying aircraft engineers on areas of concern.
A*STAR has developed new algorithms and improved on existing algorithms to detect a wider range of airplane defects with greater efficiency, accuracy and speed, while requiring less samples to train the AI model. Currently, the enhanced AI technology can accurately detect more than 20 different types of defects, including rare defect types with limited training samples.
Complementing the AI technology for defect detection are autonomous robots equipped with 3D lidar obstacle detection, accurate aircraft localisation and path planning capabilities. The combination of low- and high-reaching robots can scan for defects across multiple parts of aircraft engines.