This page intentionally left blank.
1. INTRODUCTION
The Tra.c Alert and Collision Avoidance System (TCAS), currently mandated on all large trans-port aircraft, has been shown to signi.cantly reduce the risk of mid-air collision. TCAS uses an on-board beacon radar to monitor the local air tra.c and logic to determine when to alert pilots to potential near mid-air collision (NMAC). If deemed necessary to prevent collision, TCAS will issue a resolution advisory to the pilots to climb or descend at a particular rate.
The logic used to determine when to issue an alert and which advisory to issue was the result of decades of development. The logic is speci.ed using pseudocode that has been validated through rigorous simulation studies to ensure that the system provides a high degree of safety while remaining operationally acceptable. Due to the complexity of the pseudocode, it is di.cult to modify the logic. Much of the logic is likely to require re-engineering to accommodate new surveillance systems and next-generation air tra.c control procedures.
This report explores a new approach to developing collision avoidance logic that has the potential to signi.cantly improve safety while reducing the rate of unnecessary alerts. The approach involves leveraging recent algorithmic advances and modern computing power to optimize the logic based on probabilistic models of aircraft behavior and performance metrics. The probabilistic models can be modi.ed to accommodate the anticipated evolution of the airspace, and the logic may be re-optimized as necessary with little development e.ort.
1.1 DESIGN CONSIDERATIONS
The following design considerations guided the development of the proposed approach.
.
Safety performance. The primary objective of a collision avoidance system is to increase safety. Because .ight tests can only test the system in relatively few encounter situations, much of the analysis required to prove the safety of the system must be done in simulation using an encounter model. During the development of TCAS, the NMAC rate was used as a safety performance metric. Historically, NMACs have been de.ned to occur when two aircraft come within 500ft horizontally and 100ft vertically. One of the primary goals of this research is to develop a system that has an NMAC rate lower than that of the existing version of TCAS. As will be discussed later, there are other ways to assess the safety of a collision avoidance system other than estimating NMAC rate.
.
Operational performance. A collision avoidance system should not interfere with normal, safe .ight operations. Hence, the system should avoid issuing unnecessary alerts. An excessive alert rate can result in decreased e.ciency, collision with secondary aircraft, and pilot non-compliance. In order to reliably prevent collision, the system will need to alert in situations where it might later be found unnecessary. Because it is impossible to perfectly predict how an encounter will evolve, the system will need to act conservatively. Carefully balancing the operational considerations with safety considerations is an important part of developing a
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址:Robust Airborne Collision Avoidance through Dynamic Programm(5)