Mobility

E-scooter Simulation Vision Zero Project with eMats 

Tackling safety concerns surrounding electric micro-mobility, particularly e-scooters. AI algorithms were developed to identify risky behaviors like drunk driving and tandem riding, by using a combination of VR simulations and real-world data.

​

The LSTM model showed high accuracy in detecting these behaviors:

- 86% for drunk driving

- 93% for tandem riding

​

A prototype dashboard was also created to monitor e-scooter maintenance needs. This research provided valuable tools for enhancing urban mobility safety and offers insights for stakeholders, contributing to Vision Zero goals. Future work involves model refinement, expanded data collection, and integration of advanced neural networks. 

Mobile sensing platform

"Speed tracker" is developed by WSP and validated by FellowBot. With on-vehicle Radar, GPS and LED display we can show message such as "slow down" alternatively realtime speed of the cars following the sensing vehicle, the aim is to regulate the speeding with "social nudging". We manage to reduce 1% of the speed i.e. around 1km/h in two separate road test in Stockholm.

This prototype called speed tracker is with TRL 7 (test in a real environment) and can easily be mounted to the vehicle and use as a sensing platform for speeding monitoring and general data collection (data as speed, acceleration/deceleration, position in both lateral and longitudinal). We can also use the platform to test the effects of different messages communicating between vehicles.