The project has successfully achieved the preliminary version of the SmaCS system—the TRL3 proof-of-concept design— which was presented to and evaluated by the Topic Manager during the Preliminary Design Review (PDR). Thus, functioning principles of the machine-learning algorithm were detailed and demonstrated with synthetic and real images recorded from the cabin mock-ups, performing several activities with the prototype, addressing important features of the system. This version will be taken as the basis for the subsequent development stage, which will iterate over it to attain the TRL4 version.

During the last months, the project has deployed its innovating method by mixing real data and synthetic data from 3D simulation. For the real data, real participants have been involved to perform data capture sessions over material that will be used to train the final system, aiming at having the best coverage of real situations.

The preliminary system includes the development of 3D simulation and data used to complete the learning of the machine. As a result, the AI will be able to evolve to multiple cabin configurations and additional scenarios were simulated to enhance the data provision. Some outcomes of the preliminary system will provide new field of application open for AI embedded in aviation for passengers’ safety and the improvement of cabin crew efficiency, reducing turnaround time and saving money for airliner.

SmaCS will target now the next phase over the development of the TRL4 SmaCS system focusing on training an enhanced version of the data collection, providing all the characteristics required to validate the system in the laboratory and developing TRL4 versions of the sensing camera and AI recognition platform designs. The major challenges achieving TLR4 include the quality of data collected for the generation of the tailored training data.