Smartphone Sensonomies: Crowdsourcing in-car mobile phone sensor data

Smartphones are now equipped with multiple high quality sensors (accelerometers, GPS receivers etc.), utilise high speed always-on data connections, and are habitually carried by most people. Exploiting this recent phenomenon, the aim of this project is threefold: firstly to investigate how events passively collected from large numbers of individual in-vehicle smartphones can be synthesised to provide accurate models of significant traffic-related issues (dangerous road sections, delays, surface degradation etc.), secondly to determine if this mechanism can provide high-quality data for transport-related research, and thirdly to perform initial evaluations on how this data can effectively and safely influence driver behaviour.

The project will have a number of outcomes. In the short term will be a mechanism for providing accurate and timely data on road conditions (hazards and degradations) which will be of great value to the Roads’ Authorities. The longer-term outcomes will be:

  1. A mechanism for evaluating driver behaviour. Current mechanisms rely on user reporting, and tacit recording represents a huge advance in accuracy and reliability.
  2. A system for accurately and reliably capturing fine-grained information about the road network for transport-related research.
  3. A framework for passively collecting information from drivers and then informing them of potential hazards.

It is anticipated that all three of these areas will be further explored through RCUK grant applications. This work will provide the necessary proof-of-concept and baseline data to convince the funding bodies that such research is feasible and won’t fall at the first fence.

There are also further outcomes that we would wish to explore in the future, such as how to similarly capture such data from cyclists relating to interventions (incidents short of an accident which currently go unreported), and how such data could be used to motivate both safer and eco-efficient driving.

Mark Dunlop, Marc Roper, Mark Elliot and Neil Ferguson

Image: