This project is a follow up project to two projects that implemented and validated the paper by D. Segal, A. Bar-Gill and N. Shimkin – “Reaching Airstrip in Engine Cutoff Emergency – Optimally Accounting for Intense Crosswinds and for In-Plane and Terrain-Evolving Obstacles”.
The system includes a real-time software which produce instructions on how and where to land to an airplane with an engine cut off (engine that have stopped working), according to the airplane parameters and state.
Using an optimization algorithm, the system finds the optimal landing site given certain flight conditions and guides the pilot according to the next landmark and optimal speed at any given moment.
The existed system was tested in real flights and although it correctly computed landing trajectory, real time experiment didn’t converge due to algorithm programming inefficiencies and hence can’t be used in real flights.
In our project we went through the research and optimization of the program following the quest to decrease the runtime of the system and provide better performance in real time.
We checked our optimization on real-flights data that was recorded during previous project experiments and tried to get better result with each improvement.
We got a major process and great result ending by improving the average and maximum runtime of the algorithm by x4-x5.
A really important result was the upper bound improvement, because we can predict that even in the worst case, we will get x4-x5 better performance than what measured in the real experiment. This was a significant result that could possibly bring us closer to a real-time working system.
This achievement was our project main goal – decrease the running latency and make the program go faster and in much more efficient way.