For some applications, you may need to understand how the car is being parked with respect to the road. What if you want to understand this only leveraging on accelerations? Let’s find out how.

This is a joint work with Alessio Luciani and prof. Emanuele Panizzi in the Gamification Lab.

Implicit interaction

The first step is detecting when the user is parking or un-parking the car. We do this using Bluetooth, but this can be done using different ways (in fact, we’re working on a solution that doesn’t need Bluetooth anymore).

Many modern cars have Bluetooth enabled “infotainment” (“information” + “entertainment”) systems. Usually, this BT connection is used to stream content from the smartphone (e.g., Spotify / YouTube) or handle calls.

When the Bluetooth is connected, we assume that the person is approaching the car (un-parking). When the Bluetooth is disconnected, we assume that the person has parked the car. We use this information with GPS and the accelerometer to understand where the vehicle is parked and when.

How is the car parked?

To understand how the car is parked, we developed a machine learning classifier with three classes: “parallel”, “perpendicular”, “angle”.

We do a bit of pre-processing and normalization, and then we use the classifier to infer the kind of parking.

While we have a small set of samples (due to the pandemic), we achieved a good result:

The full paper

If you want to know more, you can find the full paper published here: https://doi.org/10.1007/978-3-030-85613-7_21

Bassetti E., Luciani A., Panizzi E. (2021) ML Classification of Car Parking with Implicit Interaction on the Driver’s Smartphone. In: Ardito C. et al. (eds) Human-Computer Interaction – INTERACT 2021. INTERACT 2021. Lecture Notes in Computer Science, vol 12934. Springer, Cham. https://doi.org/10.1007/978-3-030-85613-7_21