Situation-Adaptive Autonomy for Time-Critical Takeoff Decisions

Abstract

This paper presents a probability model for analyzing risks in rejected takeofis (RTO) of aircraft. The human pilot usually bears full responsibility for deciding whether to continue the climbout (Go) or to abort the takeoff (No Go) upon an engine failure in the takeoff run. Making a correct Go/No Go decision is not an easy task under time-criticality, especially with uncertain information, inappropriate RTOs which end with overrun accidents are reported annually. Discussions on whether to automate the Go/No Go decision-making system, to avoid unnecessary RTOs made by the human pilot, have taken place. This paper proves that Go/No Go decision-making should neither be fully automated nor be left always to the pilot, and that authority for Go/No Go decision-making must be traded between the human pilot and the automated system in a situation-adaptive manner. This paper also proves that the human-interface of conventional warning systems is not effective for reducing RTO accidents, and that a new design of human-interface systems is necessary.