Autonomy in Urban Air Mobility (UAM) and UAS Traffic Management (UTM)
This research focuses on both vehicle autonomy and airspace autonomy to achieve operation safety, efficiency, and scalability for urban air mobility (UAM) and UAS traffic management (UTM), which includes arrival management, computational guidance, separation assurance, conflict resolution, autonomous air traffic control, strategic flow management and fleet dispatch.
Autonomous Drone Racing (ADR)
We are establishing both software and hardware platforms to enable high speed autonomous drone racing (ADR) in indoor dynamic environments. Our focus is on real-time learning-based perception and decision making algorithms in highly dynamic and uncertain environments.
Stochastic Models for Multiple Constrained Resources Air Traffic Flow Management
Our team is building stochastic models for sequential decision making under uncertainty in the newest Federal Aviation Administration (FAA) air traffic flow management tools.
Deep Reinforcement Learning (DRL) and Multi-Agent Reinforcement Learning (MARL)
We are creating models and algorithms to beat the state-of-the-art performance in deep reinforcement learning (DRL) and multi-agent reinforcement learning (MARL).
Towards an Intelligent Low-Altitude UAS Traffic Management System (NSF)
The research provided theoretical foundations to support UAS pre-departure traffic coordination and en route traffic management in low-altitude airspace.
Urban Air Mobility (A^3 by Airbus)
In order to enable on demand air transportation in an urban setting, our research group was working with Airbus A^3 teams on new concept of operations, modeling and simulation, and algorithm design and analysis to provide safe, efficient, sustainable and intelligent solutions for urban on-demand air transportation.
Passenger Direct Share Forecast (FAA)
The objective of this FAA project was to build machine learning based predictive models to forecast the direct passenger percentages for all the origin-destination airport pairs in the United States. The newly developed model is expected to replace the current FAA forecasting model.
Aviation Weather Impact on Flight Operations (Collins Aerospace)
Our group was working with Rockwell Collins Advanced Technology Center to study the flight en route time variation under different convective weather events. Machine learning algorithms are being developed and analyzed by mining the large-scale nationwide flight data sets (AOTP) and meteorology data sets (ASOS, HRRR, NEXRAD). The resulted predictive model is expected to provide estimation for aircraft en route time before departure given weather forecast, which is critical for decision makings in both airline operations and air traffic management.
General Aviation Safety Analysis (FAA)
The objective of this FAA PEGASAS project was to analyze aircraft accidents and incidents that occurred at or near airports and to identify actual or potential airport risks and hotspots related to those accidents and incidents. Accidents and incidents were grouped into categories based on contributing factors and broken down by airport type. The results will provide insight into eliminating or mitigating risk factors that result in incidents and accidents.