Data Science in Pilot Performance Assessment - QC-211

Preferred Disciplines: Master's student in Artificial Intelligence / Machine Learning / Data Science / Recommendation system / Signal processing / Data Visualization
Project length: 2 years (4 units)
Approx. start date: As soon as possible
Location: Montreal, QC
No. of Positions: 1
Preferences: None
Company: CAE Inc

About Company:

CAE is a global leader in training for the civil aviation, defence and security, and healthcare markets. Backed by a 70-year record of industry firsts, we continue to help define global training standards with our innovative virtual-to-live training solutions to make flying safer, maintain defence force readiness and enhance patient safety.

Summary of Project:

Automatically assessing a pilot performance during a flight training session is a capability that can enhance the flight instructor during his duty. From data gathered during a flight maneuver, we are looking for a way to automatically assess pilot performance to augment instructor performance and provide objectivity during flight training assessment.

This project is currently looking for an automatic pilot assessment capability using machine learning algorithms that can inform flight instructor during a flight training session in full flight simulator. Machine learning algorithms are considered in the approach. 

Research Objectives/Sub-Objectives:

The research project objective is to determine if and how a machine learning and algorithms can be capable of:

  • Visualizing the performance of a machine/deep learning algorithms that can detect training event and assess pilot performance from a flight training session data segment that corresponds to a flight training maneuver part of a flight training session.
  • Explaining and interpreting the results of a machine/deep learning algorithms
  • Is able to recommend training task based on pilot performance
  • Can adapt the flight simulation based on pilot performance and system recommendation

Methodology:

    • Considering a dataset of labelled flight training session data segments, the algorithm should
      • Learn how to assess pilot performance on any new unlabeled data set with proper precision and accuracy metrics.
      • Offer a machine learning architecture that will be able to assess multiple flight training tasks
      • Can explain the results of a pilot performance assessment by giving proper flight parameters that have an importance into the decision-making process and communicates it to the end user that will require such explanation and understandability of a pilot assessment.
      • Can offer interpretability of a deep learning algorithm to the data scientist so he can understand the decision-making process of the algorithm and how the model will behave in a production environment
      • Offer a machine learning architecture that will be able to assess pilot performance with limited amount of labelled data
    • Data
      • A collection of flight training session data segment: composed of timeseries of hundreds of aircraft parameters and training context. Frequency of the timeseries can vary between 10 to 60 Hz. Labelled data coming from multiple aircraft simulations will be provided. Depending on the context and conditions, however, the amount of labelled data may be limited.

    Expertise and Skills Needed:

    • Artificial Intelligence / Machine Learning / Data Science / Recommendation system / Signal processing

    For more info or to apply to this applied research position, please

    1. Check your eligibility and find more information about open projects.
    2. Interested students need to get the approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform or directly to Marie-Laure de Boutray
    Program: