CUDA Spotlight: GPU-Accelerated Air Traffic Management




Our new spotlight is on Dr. Monish Tandale, a Research Scientist at Optimal Synthesis Inc. (OSI), based in Los Altos, California. This interview is part of the CUDA Spotlight Series.

NVIDIA: Monish, tell us about Optimal Synthesis.
Monish: Optimal Synthesis Inc. (OSI) is a high-technology R&D firm specializing in algorithm and software development for guidance, navigation & automatic control and signal processing disciplines. The company was founded in 1992, and I joined as a Research Scientist in 2006.

NVIDIA: Monish, tell us about Optimal Synthesis.
Monish: Optimal Synthesis Inc. (OSI) is a high-technology R&D firm specializing in algorithm and software development for guidance, navigation & automatic control and signal processing disciplines. The company was founded in 1992, and I joined as a Research Scientist in 2006.

OSI's research focus areas include Missiles and High Performance Aircraft, NextGen Air Traffic Management Systems and Signal Processing. The company's expertise includes mathematical modeling, computer simulation, analysis, estimation, automatic control and optimization for complex dynamic systems and processes. Since 2007, high-performance computing with GPUs has been one of our core competencies.

NVIDIA: Why is traffic flow management important?
Monish: The U.S. operates the busiest, largest, most complex aviation network in the world. Every day, the U.S. National Airspace System (NAS) services tens of thousands of commercial, military and general aviation aircraft safely across the country.

Air traffic flow management involves balancing air traffic demand with system capacity in the NAS and managing the traffic flow in a safe, efficient, and coordinated manner. Effective traffic flow management not only increases passenger safety, but also mitigates economic costs associated with flight cancellation and delays, and environmental costs associated with excessive fuel burn during airborne flight delays.

NVIDIA: Tell us about your work with the NASA Ames Research Center.
Monish: At any given moment, there are over 7000 aircraft flying in the skies of the U.S. This number is only going to increase, adding more complexity to air traffic flow management. The project we are working on with NASA is related to NextGen, a multi-agency initiative to improve the coordination of air traffic flow in the NAS.

Our project is focused on predicting the "4-Dimensional Trajectory (4DT)" for every aircraft in the NAS. A 4DT describes the 3-dimensional position (latitude, longitude, altitude) of aircraft at regular time intervals (time is the 4th dimension). In the paradigm of 4DT operations, all air traffic control decisions are performed based on the predicted 4DT, i.e. the predicted position of the aircraft in the future. This project resulted in the development of CARPAT™ (Computational Appliance for Rapid Prediction of Aircraft Trajectories).

NVIDIA: How does GPU computing play a role in your work?
Monish: Rapid prediction of aircraft trajectories is critical for decision making in future Trajectory Based Operations. Repeated generation of 4DTs for various operational scenarios is necessary for iterative optimization algorithms and stochastic analysis.

GPU computing allows us to exploit the parallelism in the trajectory prediction process to have extremely fast run-times. This in turn allows us to achieve real-time performance and analyze models with greater complexity, and opens up the possibility of utilizing algorithms and approaches that were earlier deemed impractical due to the computational complexity.

NVIDIA: What kind of advantages have you achieved with GPU computing?
Monish: Using CUDA and GPU computing we were able to achieve a 250X speedup over NASA’s baseline software called Future ATM Concepts Evaluation Tool (FACET). While FACET takes around 10 minutes to perform a typical 24-hour trajectory prediction for 35,000 aircraft in the NAS, the CARPAT prototype performs the same prediction in under 2.5 seconds. Also, the shallow learning curve associated with CUDA allowed us to smoothly transition from C/C++ based CPU programming to GPU programming.

NVIDIA: When you think about the future of air traffic management, what will be the biggest change from today?
Monish: Today’s air traffic management is mainly human-centric, relying on the experience of air traffic controllers and coordinators to make decisions for managing traffic flow. The approach has served us well in the past. However, with ever increasing demand for air travel, we need computer-based decision support tools to assist human operators in making their decisions.

With increasing computational capability available at our disposal, we can develop and analyze highly complex traffic flow models and run optimization algorithms to generate optimal schedules and routing of aircraft in the NAS to achieve the maximum possible efficiency.  The goal is to use automation to allow more people to fly without being inconvenienced by congestion and delayed flights.

NVIDIA: Are you working on any other projects using GPU computing?
Monish: Yes, we currently we are pursuing multiple research projects involving GPU computing, funded by various agencies such as NASA, the Missile Defense Agency and the Air Force. 

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Bio for Dr. Monish Tandale:
Monish Tandale received his M.S. (2002) and Ph.D. (2006) degrees in Aerospace Engineering from Texas A&M University, specializing in Dynamics & Control. Since 2006, he has been a Research Scientist at Optimal Synthesis Inc. At OSI, he has participated in and led several NASA and Department of Defense R&D projects in the areas of NextGen Air Traffic Management, Nonlinear Estimation, Queuing Theory, Numerical Optimization and High Performance Computing using Graphics Processing Units. He is passionate about solving computationally challenging problems in the Aerospace & Defense domain using GPUs.

Relevant Links:
Conference Publication: Tandale M. D., Wiraatmadja S., Menon P. K., Rios J. L., “High-Speed Prediction of Air Traffic for Real-Time Decision Support,” Paper Number AIAA-2011-6660, AIAA Guidance, Navigation, and Control Conference, Portland, Oregon, Aug. 8-11, 2011: http://www.aiaa.org/content.cfm?pageid=298

Presentation at the NVIDIA GPU Technology Conference (GTC) 2010, San Jose, CA: http://nvidia.fullviewmedia.com/gtc2010/0921-l-2214.html

Air Traffic Management Research at the NASA Ames Research Center: http://www.aviationsystemsdivision.arc.nasa.gov/research/index.shtml

NASA’s Future ATM Concepts Evaluation Tool (FACET):http://www.aviationsystemsdivision.arc.nasa.gov/research/modeling/facet.shtml

Website: http://sites.google.com/site/monishtandale

Contact Info:
Email: monish@optisyn.com