SBIR-STTR Award

The SSCI team proposes to develop and test the on-board Fast Online pREdiCtion of Aircraft State Trajectories (FORECAST) system, using minimum state information such as 3-D position of a threat aircraft, to generate predicted trajectories and reachable sets T seconds into the future. It will be based on a nonlinear constrained stochastic model of aircraft dynamics involving rapid maneuvering, advanced nonlinear filtering techniques, and the design of the predicted exclusion zone for the aircraft operating in the vicinity of the threat aircraft. The algorithms used to develop the FORECAST technology will include: multi-model nonlinear filtering using Interacting Multiple Models; Extended Kalman Filter; Fokker Planck Equation; and exclusion zone calculation using stochastic feedback version of the Rapidly-exploring Random Trees algorithm. In Phase I, we will test the FORECAST system on a simplified scenario simulation. The Option will include extensive testing on a higher-fidelity simulation. In Phase II, we will continue algorithm development, perform extensive simulations and flight testing at MIT''s RAVEN facility, and develop the FORECAST software toolbox. Our academic partner, Prof. Jonathan How of MIT, brings in a wealth of expertise and experience in the area of 4-D trajectory planning, autonomous UAV control, multi-agent collaboration, and advanced flight test facilities.

Benefit:
Improved capability in mid-air collisions predictions leading to fewer nuisance warning, increased user acceptance, and integration of unmanned aerial systems into the National Airspace is a key technology component for the safety of the air vehicles and its applications. Homeland Defense and law enforcement will also benefit from these technologies. Commercial applications of trajectory prediction techniques and systems exist in areas such as air traffic control and space situational awareness.

Keywords:
Unmanned Aerial Systems, Unmanned Aerial Systems, Sensors, Mid-Air Collision Avoidance Systems, Trajectories, Detect and Avoid, Sense and Avoid

The main objective of the Phase II work is to enhance the FORECAST algorithms developed in Phase I, extend them to 3D in the case of multi-rate sensors and significant wind effects, develop effective tuning procedures, and evaluate the performance of the FORECAST system on test data. In order to achieve these objectives, we plan to carry out the following tasks: (i) Extend FORECAST algorithms to 3D in the case of multi-rate sensors and significant wind effects; (ii) Develop automated algorithm tuning procedures; (iii) Evaluate the proposed algorithms and models for estimation and prediction on test data; and (iv) Develop a prototype FORECAST system and deliver it to the Navy. Under this project our academic partner is MIT (Prof. J. How). The MIT team will extend the reachable set algorithms to 3D, and flight test the FORECAST algorithms in their indoors facility.

Keywords:
Nonlinear Trajectory Prediction, Nonlinear Trajectory Prediction, Mid-Air Collision Detection, Nonli