Sonja Koesling on how crowdsourcing technology is making significant inroads into the advanced traffic management arena.
It started out like Hansel and Gretel in 2004, when Steve Coast started to leave the first breadcrumbs with his newly purchased GPS device. He plotted his journeys covered on a map, wrote the software to do so and encouraged others to join him and so OpenStreetMap (OSM) was born. Today, his vision of a freely available world map is one of the most successful crowdsourcing projects and is both the basis and inspiration for others. This also applies to traffic quality and road safety.
Crowdsourcing, a term coined by the US journalist Jeff Howe, is an interactive form of value creation by a group of people – a crowd. Crowdsourcing has become popular in the Internet era and therefore usually involves a network community in the World Wide Web. Crowdsourcing is based on the idea of group intelligence, which means that the decisions of a homogeneous group of individuals can achieve the same quality as expert opinions. Crowdsourcing also relies on the enthusiasm of the people in the crowd. But what is the motivation of the crowdsourcees?
“Crowdsourcees such as the OSM mappers draw their motivation from their personal interests,” explains Coast. “They might be cyclists who want better cycling maps for various parts of the world or walkers who want to map their routes and many more. OpenStreetMap can be seen as the sum of the personal interests of all participants who want to see them recorded on the map.”
Always mobile, always locatable
Thanks to smartphones and apps, it’s now easy to record your own GPS tracks and upload them to OpenStreetMap. “OpenStreetMap then shows the track in a Flash-based editor called Potlatch”, explains Coast. “This editor provides a view of the map data, the recorded GPS track as well as satellite images provided by Microsoft for the project. This means mappers can then photographs roads, railways or rivers along the recorded points as well as adding objects such as letter boxes, trees or anything they have noticed as well as amending objects that have already been added.”
The key to generating high-quality map data is object tagging. While commercial providers follow very formal ontologies and assign certain attributes to roads from a defined set, OpenStreetMap trusts in the wisdom of the crowd and that mistakes will be eradicated by the crowdsourcees.
Rating traffic lights
A research project in Austria shows that the wisdom of the crowd can in fact yield good results. Inspired by rating portals such as Tripadvisor, the project built an online feedback platform that allows road users to intuitively rate the traffic quality and safety of signal-controlled intersections. TrafficCheck.at is the name of the project funded by the Austrian ministry responsible for traffic, innovation and technology (bmvit). It relies on the mapping and tagging enthusiasm of its crowdsourcing contributors and harnesses this enthusiasm to rate traffic lights using a simple app.
With their smartphones, TrafficCheck contributors can track their positions on OpenStreetMap while underway and select the signal-controlled intersections they wish to rate. They enter their transport mode to provide information on whether the rating comes from the perspective of a motor vehicle driver, cyclist, pedestrians or public transport passenger. The crowdsourcing contributors can then report problems or enter a score, ranging from 1 star for an intersection regarded as bad to 4 stars for one regarded as satisfactory. The score is based on factors such as waiting time and green time, visibility, conflicts between road users and system layout.
The feedback from the crowdsourcing contributors goes directly to the municipal authorities and is incorporated into the traffic planning process and the traffic control systems. This is intended to improve the level of service of the intersections and enhance road safety. Results from initial live tests show that the crowd can provide structured and systematic results even in complex situations and can recognise weaknesses at intersections. Crowdsourcing projects such as TrafficCheck.at can also provide useful information to supplement objective criteria for managing the quality of traffic light systems.
Can crowdsourcing improve road safety?
The question is whether crowdsourcing can actually yield valid results for road safety. “I think that crowdsourcing projects such as TrafficCheck can capture subjective experience and thereby increase awareness of critical points in the road network”, says Andre Münch, Director Business Development Safety at the PTV Group. “Of course, they cannot replace the detailed evaluation of traffic accidents by the police.” So crowdsourcing can’t do everything at that point. But mapping and tagging is also an issue in the police departments.
In recent years the police have adopted GIS as an ideal alternative to paper-based accident-type maps. Instead of large maps with lots of coloured pins, software is used to link accident data directly to a digital road map. This allows the data to be analysed by the police and filtered by features. The main objective of this is to detect similar anomalies. For example, if an unusually large number of accidents occur at a specific location in combination with wet weather or icy conditions, this may indicate a problem with the road surface. If this suspicion proves to be correct, suitable measures can be taken and their effect on subsequent accident incidence can be observed.
In Germany, the police in 11 states all use PTV Euska to analyse accident data. This software was developed by the PTV Group, in cooperation with insurance company accident appraisers, under the auspices of the Accidents Commission in Germany to support the joint activities of the police and municipal authorities. This product is also planned to be rolled out internationally in the coming months.
However, Münch cautions: “If our long-term vision is the zero-accidents scenario, the reactive approach we have taken up to now is not enough. We need a paradigm shift: road safety must become an integral part of traffic planning and be anchored more strongly in the strategic planning stage.” That is only possible with closer collaboration between the police, who collect and evaluate the accident data, and the planning authorities, who plan the extensions and restructuring of the traffic infrastructure to accommodate future traffic demand. On the tools front, PTV Visum Safety can help these two organisations close ranks.
From reactive to proactive
Like PTV Euska, Visum Safety supports Black Spot Management, which means the identification and analysis of accident clusters and points with high accident rates. This allows individual accidents to be filtered by their attributes, visualised in the form of heat maps, and analysed quickly. “A special aspect of Visum Safety is that it allows users to merge accident data with traffic volumes”, notes Münch. “This enables traffic planners and safety analysts to evaluate especially critical points in the network with regard to traffic significance as well as other aspects, and to use segment colour coding and interactive tables to establish an urgency ranking.” In addition, the potential safety benefit or urgency of infrastructure measures in the susceptible network segments can be identified and evaluated in financial terms – i.e. avoidable accident costs. Experts call this approach Network Safety Management (NSM).
NSM typically includes both reactive and proactive components. Identification of critical network segments requires historical accident data, which both traffic planners and road network operators can use to proactively develop and plan potential measures. The next level is Road Safety Impact Assessment (RIA). “RIA enables a risk assessment of existing and future infrastructure”, says Münch. “Traffic planners already examine the economical and environmental impacts of planning options in the traffic models for their infrastructure projects. With RIA they can expand the scope of examination to include road safety issues.” On the tools side, this is facilitated by including simplified accident prediction models for various road types in PTV Visum Safety.
By merging this information with the traffic volumes from the traffic model, traffic planners can draw conclusions about predicted accident rates for individual roads or the entire network and study various scenarios. “In future RIA will be increasingly important, because the most effective way to prevent accidents is to predictively model accident incidence in order to choose the apparently safest planning option”, concludes Münch.