Event Information
Workshop organizers: Prof. Andrea L'Afflitto (Virginia Tech) and Prof. Gokhan Inalhan (The Sloane Institute)
Organization: IEEE CSS Technical Committee on Aerospace Control
Venue: IEEE Conference on Control Technology and Applications
Date: Sunday, August 24
Location: Westin Gaslamp Quarter, San Diego
Workshop Schedule
| Time | Session | Speaker |
|---|---|---|
| 09:00 - 10:00 | Safe Learning and Control with L1 Adaptation | Prof. N. Hovakimyan Center for Autonomous Vehicles in Air Transportation Engineering, The Grainger College of Engineering - Mechanical Science & Engineering, University of Illinois, Urbana-Champaign, USA |
| 10:00 - 10:30 | Coffee Break | — |
| 10:30 - 11:15 | Enabling Sustainable Aviation Through Advanced Data Analytics and AI | Prof. G. Inalhan The Sloane Institute, London, United Kingdom |
| 11:15 - 12:00 | Roundtable on the Future of Control and Learning | Moderator: G. Inalhan The Sloane Institute, London, United Kingdom |
| 12:00 - 13:30 | Lunch | — |
| 13:30 - 14:30 | Constrained Adaptive Control on Reproducing Kernel Hilbert Spaces with Applications to Autonomous Uncrewed Aerial Vehicles | Prof. A. L'Afflitto Grado Department of Industrial and Systems Engineering, Virginia Tech, USA |
| 14:30 - 15:30 | Robot Dynamics Learning and Control using Port-Hamiltonian Neural ODE Networks | Prof. N. Atanasov Department of Electrical and Computer Engineering, University of California, San Diego, CA, USA |
| 15:30 - 16:30 | Cooperative Variable Bandwidth Actuators Designed for Aerospace Applications | Dr. R. Hull Collins Aerospace, Kissimmee, Florida, USA |
| 16:30 - 17:00 | Roundtable on the Future of Aerospace Control from Academia and Industry Perspectives | Moderator: Prof. A. L'Afflitto Grado Department of Industrial and Systems Engineering, Virginia Tech, USA |
Presentation Abstracts
Safe Learning and Control with L1 Adaptation
Prof. N. Hovakimyan
Learning-based control paradigms have seen many success stories with autonomous systems and robots in recent years. However, as these robots prepare to enter the real world, operating safely in the presence of imperfect model knowledge and external disturbances is going to be vital to ensure mission success. We introduce a class of distributionally robust adaptive control architectures that ensure robustness to distribution shifts and enable the development of certificates for V&V of learning-enabled systems. An overview of different projects at our lab that build upon this framework will be demonstrated to show different applications.
Enabling Sustainable Aviation Through Advanced Data Analytics and AI
Prof. G. Inalhan
This talk showcases the state-of-art in digital twin performance modelling and flight planning towards performance-emission-climate impact optimized operations. Specifically, physics-informed deep learning based aircraft performance modelling techniques are introduced to design high precision aircraft models. These models not only enable optimal fuel loading, but also allow the operators to utilize wind-optimized trajectories while reducing emissions and potential climate impact. As such, the newly developed tracking-information based business analytics allow the operators to benchmark their performance to other operators in terms of origin-destination pairs across numerous metrics such as fuel usage, operation costs and climate impact. Using such benchmarks, the aircraft operations can be further optimized to achieve the desired balance towards cost and emissions. The presented technologies and their effectiveness are demonstrated with data from real-world operations.
Constrained Adaptive Control on Reproducing Kernel Hilbert Spaces with Applications to Autonomous Uncrewed Aerial Vehicles
Prof. A. L'Afflitto
The problem of modeling and countering the effect of uncertainties affecting nonlinear dynamical systems has been at the core of control theory. Although this problem has been addressed by a wide range of control techniques, it is still considered open since all existing solutions address specific plant models or specific forms of uncertainties. Recently, the presenter and his collaborators created robust adaptive control techniques that merely assume uncertainties and disturbances to lie in some user-defined reproducing Hilbert space. Applied to autonomous uncrewed aerial vehicles, these results have proven to provide lower control efforts and higher tracking precision in challenging problems such as those involving the transportation of unknown, unsteady payloads.
Robot Dynamics Learning and Control using Port-Hamiltonian Neural ODE Networks
Prof. N. Atanasov
The problem of modeling and countering the effect of uncertainties affecting nonlinear dynamical systems has been at the core of control theory. Although this problem has been addressed by a wide range of control techniques, it is still considered open since all existing solutions address specific plant models or specific forms of uncertainties. Recently, the presenter and his collaborators created robust adaptive control techniques that merely assume uncertainties and disturbances to lie in some user-defined reproducing Hilbert space. Applied to autonomous uncrewed aerial vehicles, these results have proven to provide lower control efforts and higher tracking precision in challenging problems such as those involving the transportation of unknown, unsteady payloads.
Cooperative Variable Bandwidth Actuators Designed for Aerospace Applications
Dr. R. Hull
Control surface actuators are a critical component of the control system design for any aerospace vehicle. Typical applications include fins, canards, ailerons, stabilizers, and rudder actuator controllers on aircraft, missiles, or guided projectiles. Many different configurations exist in which combinations of two, three, four, or more control surfaces are used collectively or differentially to control roll, pitch and yaw motions of the air vehicle. When each actuator is controlled independently and driven by position and/or speed commands from the external autopilot, it is difficult to control desired airframe response in pitch, roll and yaw precisely. Due to parameter variations and different aerodynamic loading on each surface, unwanted perturbations airframe response may occur. Another issue is that actuators designed for a fixed bandwidth to meet worst case loading response characteristics, may be inefficient in use of energy and susceptible to noise in less stressing regions of flight. The proposed solution alleviates both problems by combining cooperative control methods with extended high gain observers in the design of Cooperative Variable Bandwidth Electric Motor Actuator Controllers. This presentation describes the design of a cooperative controller that receives roll, pitch and yaw commands from the autopilot, motor position and current feedbacks from each of the multiple actuators, and then provides optimum motor voltage commands to drive each of the actuators. Embedded high gain observers are able to estimate both the loading torque and torque disturbances on each motor, and compensate to eliminate unwanted airframe perturbations. This approach allows for the bandwidth (or response time) of each axis (pitch, roll, yaw) to be controlled independently, and even to be modified during flight by commands from the external autopilot to achieve energy efficiency. Simulation results under hypothetical conditions will be presented to demonstrate these advantages.