A-PiMod (Applying Pilot Models for Safer Aircraft) project funded by the European Union is contributing to increasing the safety of air transport through the development of a new, multimodal and adaptive cockpit architecture – the cockpit of the future.
The use of automation in the cockpit is currently generally dependent on technical factors. The condition of the pilots is not taken into consideration in the majority of cases. This can lead to the cockpit automation acting in a way that is unintelligible for the pilot. As accident investigations have shown, problems such as these in the interface between human and automation contribute to 60 to 80% of all accidents.
The objective of A-PiMod
The A-PiMod project aims to develop a cockpit architecture which can be viewed as a third member of the cockpit team alongside the two pilots. A-PiMod is intended to improve cooperation between humans and machines and contribute to reducing the number of errors in the cockpit and limiting their impacts.
The cockpit architecture
The cockpit architecture developed combines multimodal interaction technologies, a pilot model and a real-time risk assessment. It supports the adaptation of the flight plan to the wishes of the pilots and the requirements of the situation, taking into account current conditions, division of tasks between pilots and automation and the modality of interaction.
Adaptation of the flight plan
In order to guarantee a safe flight, the systems are permanently monitoring and evaluating aircraft condition, environmental conditions and the condition of the pilots. Possible dangers can thus be recognised. Information about risks identified are shown on a newly developed management display. Based on this, the pilots are able to select possible flight changes via the display and to have the system re-evaluate the risks involved.
Adaptation of the division of tasks
The cockpit architecture developed supports the dynamic division of duties between the cockpit crew and the automation. To achieve this, a component first identifies the tasks to be carried out in the current situation. It then determines all possible allocations of tasks and the resultant probabilities for error. One important criterion for determining the error probability is the condition of the crew. Based on eye-gaze, gestures and inputs by the pilots, conclusions are drawn about their current intentions, their situational awareness and their workload. Pilots can modify the allocation of tasks using the management display.
Adaptation of the interaction
To make cooperation between the pilots and systems in the cockpit easier, voice recognition software and touchscreens are added to the conventional modalities of interaction. This has been implemented using a multimodal navigation system as an example. In addition, a purely passive system for capturing gesture and eye movements has also been developed.
To what extent A-PiMod achieves the expected objectives was regularly assessed. 13 pilots tested the systems in a DLR cockpit simulator as part of the project. In general, the participants confirmed the positive contribution that A-PiMod can make to increasing safety by reducing errors in the cockpit.
The modules developed in A-PiMod may in future also support flight instructors in flight simulators. A training tool is able to show them innovative performance measurement information by means of a tablet and can deliver innovative findings with regard to eye movement, pilot condition, aircraft condition and possible dangers. In addition, the training tool permits the flight instructor to evaluate the skills of pilots.
The research leading to A-PiMod results has received funding from the European Commission Seventh Framework Programme (FP7/2007-2013) under contract number: 605141 Project A-PiMod.
Deutsches Zentrum für Luft- und Raumfahrt e.V. (Lead)
Brno University of Technology
NLR (Stichting Nationaal Lucht- en Ruimtevaartlaboratorium)
Symbio Concepts & Products
Trinity College Dublin