Advanced Air Mobility

The imminent arrival of Advanced Air Mobility (AAM) signals the future of air transportation. In our lab, we focus on researching fundamental flight technologies essential for unlocking the potential of future air mobility. 

Operational Domain Design (ODD) for Advanced Air Mobility (AAM)

In this research, we aim to devise a systematic method for defining an Operational Domain Design (ODD) for automated flight systems of future aircraft. We build upon recent advancements in the automotive industry in this area and leverage the expertise in flight operation and technology contributed by Sunny Yun who is a Ph.D. student in our lab and an active pilot with Asiana Airlines, flying A350.

The ODD, which defines the operating environment of an automated system or a function, has been adopted widely for autonomous driving cars for multiple purposes, for instance, to share safety information clearly between stakeholders, to derive engineering requirements for automated systems and/or features under development, and to carry out scenario-based verification of the automated systems. In order to construct an ODD, one needs to develop a well established taxonomy based on a set of attributes representing different aspects of operating conditions of an automated system/feature. In recent years, significant effort has been made in an automotive industry to come up with an effective ODD taxonomy, together with some recommendations on how to format an ODD under such taxonomy. This led to the release of several standard documents including ISO 34503 and BSI PAS 1883. 

On the other hand, as the era of AAM dawns, there has been a burgeoning interest to introduce an automated pilot assistance system to the next generation aircraft that will help reduce the pilot's workload and enhance the flight operational safety by eliminating human errors. Indeed, autonomous flight is considered as a distinguishing feature of AAM aircraft compared to the current existing aircraft, which will have a dramatic impact on how the future aircraft would fly and operate. To bring autonomously flying aircraft into the real world, substantial effort has been invested, not only by engineering  experts but also by regulatory bodies, in the aviation industry to develop various automated flight systems. While these systems would have their own designed operating conditions and environment, a consensus on how to best represent them as an ODD is yet to be reached, unlike in the automotive industry.  To this end, our research focuses on laying the foundations of ODD construction for automated flight systems by thoroughly considering real-world flight operations and technologies.

Vision-based autonomous landing system of a fixed-wing aircraft

In this research, we aim to develop a vision-based autonomous landing system for a fixed-wing aircraft that can be deployed in the real-world. Such a system could be especially useful for the future fixed-wing type aircraft, such as Regional Air Mobility (RAM) aircraft.

In the field of aviation, the emerging concept of Advanced Air Mobility (AAM) has spurred the demand for autonomous aircraft capable of flying with limited or no human intervention. This is because autonomous flight technologies are envisaged to mitigate the risks associated with human errors by pilots and to facilitate quicker reactions to adverse flight conditions than human pilots. In particular, it is well-known that landing is the most dangerous phase of flight, and efforts have been made to develop an automated system that could assist pilots during landing and improve landing safety.

However, current landing assistive systems, such as Instrument Landing System (ILS), require ground equipment and a-priori knowledge of runways. To overcome these limitations, we propose using an onboard camera that looks ahead of the aircraft to identify a runway from camera images when it comes into sight. A deep-learning based segmentation method may be used to recognize runway when approaching the runway from mid-air, and landing guidance command is generated based on the estimated relative pose of the aircraft with respect to the runway that is obtained by fusing the image with other sensor measurements such as an Inertial Measurement Unit (IMU). Hence, the method does not rely on any ground equipment and runway information. Furthermore, this method may be employed to safely execute emergency landings in areas other than runways by identifying a safe emergency landing spot using onboard vision and issuing the necessary landing guidance commands.

Fail-safe strategies for AAM aircraft

For aircraft, emphasizing the importance of safety is never excessive. While safety considerations of an aircraft can encompass various factors depending on the context, our research focuses on developing critical methods for situations where a subset of flight controls on a vertical take-off and landing (VTOL) aircraft malfunctions during flight, enabling emergency flight operations.

We approach this problem from two different angles. The first involves carefully designing an aircraft to maintain flight controllability even when a certain subset of its flight controls fails. The other approach is to implement a dedicated flight controller capable of ensuring safe emergency flight despite the loss of flight controllability due to failed control effectors. Ultimately, our plan is to integrate both design and control approaches, optimizing the fail-safe functionality of an aircraft.

As a starting point, we considered a multirotor aircraft susceptible to partial rotor failure and proposed a control-centric design optimization approach. This method optimizes aircraft design parameters to ensure the aircraft's survivability in the event of rotor failure up to a certain number, irrespective of the combination. We continue to explore the design space of multirotor aircraft and seek other practical means to cope with rotor failure. Eventually, we plan to expand our research outcomes to a broader class of VTOL aircraft.