Hartmut Helmke

Scheduling algorithms for ATM applications — Tools and toys

Planning, particularly scheduling of limited resources is one of the main tasks of ATM. More than 10 papers of ATM seminar 2009 covered scheduling. None of these papers, however, generalizes its results so that they are suitable for different ATM scheduling problems. In this paper we extend common heuristics. We show that the A∗ algorithm should be preferred to Branch&Bound (B&B). A∗ achieves the same quality results, but with a runtime reduction of 50 to 70 %. The presented Take-Select heuristic with or without iterations in combination with a base algorithm like A∗ or B&B enables the optimization of some hundreds of elements. It enables the needed trade-off between objective function quality and runtime performance by adjusting the window size. We demonstrate the performance of Greedy, Tabu-Search, Branch&Bound, A∗ and Take-Select with different window sizes by applying them for ground vehicle and aircraft scheduling. We use our algorithms to optimize the landing sequence at an airport with very closely spaced parallel runways and show that these algorithms are also able to simultaneously optimize the usage of multiple resources.

Reducing controller workload with automatic speech recognition

Air traffic controllers normally manage all aircraft information with flight strips. These strips contain static information about each flight such as call sign or weight category. Additionally, all clearances regarding altitude, speed, and direction are noted by the controller. Historically paper flight strips were in operation, but modern controller working positions use electronic flight strips or electronic aircraft labels. However, independent from the type, considerable controller effort is needed to manually maintain strip information consistent with commands given to the aircraft. Automatic Speech Recognition (ASR) is a solution which requires no additional work from the controller to maintain radar label information. The Assistant Based Speech Recognizer developed by DLR and Saarland University enables command error rates below 2%. Validation trials with controllers from Germany and Austria showed that workload reduction by a factor of three for label maintenance is possible.

Optimal mixed-mode runway — Mixed-integer programming for ATC scheduling

⇁ MIP provides dual solution quality information which can be used to evaluate the quality (i.e. optimality) of solutions. ⇁ Compacter sequences and therefore increased throughput is possible with a holistic approach. ⇁ ADCO more fits for the 70:30 usage of the runway (70% ARR). Usage of existing AMAN/DMAN investments is possible. ⇁ The holistic approaches are applicable for full mixed-mode operation (60:40 to 50:50 to 40:60). New tools may be needed. ⇁ ADCO implicitly generates more buffer, holistic approaches explicitly model the buffers. ⇁ All three approaches, however, increase the throughput and reduce controller workload compared to a master slave scenario.

Optimal mixed-mode runway scheduling — Mixed-integer programming for ATC scheduling

On mixed-mode runways arriving and departing aircraft share the same runway. Whereas arriving aircraft have rather tight constraints with respect to their touch down time departing aircraft can wait on-ground. Sufficiently large gaps have to be generated in between two different aircraft. We therefore have to schedule arriving and departing aircraft in an integrated fashion taking a diverse set of requirements, e.g., minimum separation, tight touch down windows, etc. into account. We present an integrated approach to mixed-mode runway scheduling using mixed-integer programming techniques. These techniques for solving optimization problems provides us with a quality guarantee of the computed solution as well as flexibility to incorporate additional constraints and in particular to use non-monotone objective functions. The resulting mixed-integer program can be solved fast using modern solvers. We also provide a brief comparison to other approaches using a real-world simulation and highlight advantages and disadvantages of our approach.

First results of coupling of ATM planning systems with different time horizons

With the implementation of Total Airport Management we expect that the pre-tactical airport planning for a 24h period will be tool supported in future. Usually decisions made and implemented in those pre-tactical phase influence the tactical planning. The paper at hand describes the trials performed with coupled ATM planning systems for the mentioned pre-tactical planning and the tactical controller assistance systems. It starts with a brief description of the pre-tactical planning system prototype from DLR and our research prototypes for arrival and departure management. The conceptual idea of coordination profiles to couple both kinds of planning systems are explained in short before selecting two coordination profiles for the trials. The result confirms our expectation: With the coupling the traffic flows entering the tactical planning are smoothed.

Context-based recognition network adaptation for improving on-line ASR in Air Traffic Control

This paper presents an approach for incorporating situational context information into an on-line Automatic Speech Recognition (ASR) component of an Air Traffic Control (ATC) assistance system to improve recognition performance. Context information is treated as prior information to reduce the search space for recognition. It is integrated in the ASR pipeline by continually updating the recognition network. This is achieved by automatically adapting the underlying grammar whenever new situational knowledge becomes available. The context-dependent recognition network is then re-created and substituted for recognition based on these context-dependent grammars. As a result, the recognizers search space is constantly being limited to that subset of hypotheses that are deemed plausible in the current situation. Since recognition and adaptation tasks can be easily performed by two separate parallel processes, on-line capabilities of the system are maintained, and response times do not increase as a result of context integration. Experiments conducted on about two hours of ATC data show a reduction in command error rate by a factor of three when context is used.