A behavioural planning capability for autonomous aircraft

The objective is to expand on current collision avoidance path planning frameworks in a two-fold manner, improving prediction as well as decision-making components. The first task is to improve prediction models and generate maneuver bands for air traffic circuit navigation. The second task is to then improve the decision-making framework to select the pseudo-optimal maneuver, i.e. the maneuver that is safe and most similar to one performed by a human pilot in the same situation. This will most likely require the development of a novel metric that scores each maneuver allowed by ownship performance limits such that maneuvers can be ranked and the pseudo-optimal maneuver can be selected. The work is expected to leverage NASA Langley’s DAIDALUS framework as a starting point, and may involve additional frameworks (e.g. ACAS-Xu, TCAS II-7.1) as the project evolves. Implementing an improved behavioural planner that expands on state-of-the-art frameworks will allow Ribbit to facilitate safe and expedient flight operations out of uncontrolled airports.

Faculty Supervisor:

Krzysztof Czarnecki

Student:

Partner:

Ribbit

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

University of Waterloo

Program: