Quality / Observation

Estimating time-varying O/D information on the base of detector pulse data and FCD measurements

Phase lengths in coordinated areas should depend on the number of cars driving from the dif-ferent accesses into the different egresses on each intersection. Most phase length and split optimization algorithms put a strong effort on the main, mostly through movements, whereas outflows (turn movements) and inflows (from secondary approaches) are difficult to handle. Therefore

  • Precise and time-varying O/D information inside a coordinated area is important for phase length and split adjustment of the intersections within the observed area.
  • Travel time prediction in case of disturbances becomes more reliable as the usual O/D pattern of the involved vehicles is known.
  • Routing suggestions can base on more precise information.

This paper presents a method that combines

  • detailed microscopic car following by pulse data where detectors are available

with

  • rather sparse FCD data each time when available

in order to obtain time-varying O/D estimates and travel time measurements.

Download the paper from here: Detroit  [PDF, 901 KB]


Do my loop detectors count correctly? A set of functions for detector plausibility testing

Loop detectors are wide-spread and relatively cheap detecting devices. With today's traffic communication revolution, high resolution detector data is available on a central traffic management level. High resolution detector data consist of detector slopes, also called pulse data.

There is an initial and continuous need for checking the detectors for correct data as all kinds of disturbances may add erroneous information to the data.
This paper proposes pulse data checking and interval data checking with optional data replacement in order to guarantee a continuous data flow even if detectors do not deliver the expected data quality.

  • - Raw detector data checking analyses rising and falling slopes of detector signals.
  • - Cumulative data checking compares interval values to reference curves.

Cumulative data checking needs less computational effort, but need more parameterization effort than raw detector data checking. Both checking principles are applied to different systems in Switzerland since about five years.

Download the paper from here: Bordeaux_ITS-1560  [PDF, 612 KB]


Measuring public transport priority quality with special attention to competing calls

Differences in traffic quality and traffic control quality can easily be felt when traveling to different cities. Why are these differences? One reason is probably due to cultural differ-ences in driving behaviour. But one important reason – that can be measured – is the quality of the control applied at the individual intersections. Intersection control quality is usually not measured systematically. Thus there is no general understanding about what type of traf-fic control leads to what quality under what circumstances. This article's main concern is public transport priority quality. But public transport priority affects directly individual traffic quality. Therefore a second concern is its impact on individual traffic. The article introduces a formalism that has been applied in Basel, Switzerland, a city with a high public transport vehicle density, composed of trams and busses, and of cars, bicycles and pedestrians. It further describes how the quality values for public transport and individual traffic can be measured and displayed in diagrams that are easy to understand. Finally it introduces addi-tional quality measurements for control operators that can be used in case of insufficient quality to find the parts in the control logic to be improved.

Download the paper from here: Melbourne_ITS-EU-TP0071  [PDF, 643 KB]


Calculating time loss by impulse detector data for transport quality measurement

Differences in traffic quality and traffic control quality can easily be felt when traveling to different cities. At the former World Congress an approach for systematically measuring the quality of the control applied at individual intersections has been proposed under the main focus of public transport priority quality. As public transport priority affects directly individual traffic quality, individual traffic control quality needs to be measured as well. In the mentioned approach individual traffic quality has been estimated under the assumption of sparse detector data availability for individual traffic. This paper now focuses on detection possibilities for measuring or estimating loss time for individual traffic, therefore giving a more solid base for systematic traffic quality and traffic control quality measurement for public transport and individual traffic and its comparison. Measuring travel time needs sufficient detectors with sufficient detection precision. Errors like cross-talking, opposite direction traffic measuring, non-observable vehicle inflow and outflow need to be discovered and compensated for. The paper introduces a method that bases on car platoon detection by signal pattern correlation on subsequent detectors. Platoon detection is more precise than individual car detection; therefore platoon detection can also be used for periodical re-calibration purposes of the estimation algorithm.

Download the paper from here: Montreal_EU-TP0759  [PDF, 487 KB]


Finding traffic quality measures with signal change data only

Differences in traffic quality perception can easily be felt when traveling to different cities. At former World Congresses an approach for systematically measuring the quality of the control applied at individual intersections has been proposed under the main focus of public transport priority quality. A follow-up paper was focusing on detection possibilities for measuring or estimating loss time for individual traffic. Now we would like to focus on pedestrians and bicycles without detection and the travelling quality to be expected. The paper introduces a method that bases on statistical sources that generate distributions for pedestrians and bicycles that enable quality evaluation with signal group change data only. The paper puts a special focus on pedestrians and bicycles crossing several lights in a row.

Download the paper from here: Singapore_EU-TP1706  [PDF, 1.00 MB]