A smart e-scooter with embedded estimation of rear vehicle trajectories for rider protection

Abstract

This paper develops an active sensing and estimation system for protecting the rider of an e-scooter from car-scooter collisions. The objective is to track the trajectories of cars behind the e-scooter and predict any real-time danger of car-scooter collision. If the danger of a collision is predicted, then a loud car-horn-like audio warning is sounded to alert the car driver to the presence of the scooter. A low-cost (∼$100) single-beam laser sensor is chosen for measuring the positions of cars behind the scooter. The sensor is mounted on a stepper motor and the region behind the scooter is scanned to detect vehicles. Once a vehicle is detected, its trajectory is tracked in real-time by using feedback control to focus the orientation of the laser sensor such as to make measurements of the right front corner of the vehicle. A nonlinear vehicle model and a nonlinear observer are used to estimate the trajectory variables of the tracked car. The estimated states are used in a receding horizon controller that controls the real-time position of the laser sensor to focus on the vehicle. The developed system is implemented on a Ninebot e-scooter platform. Extensive experiments conducted with multiple vehicle maneuvers show that the closed-loop system is able to accurately track vehicle trajectories and provide audio alerts to prevent collisions. This paper constitutes the first-ever development of active rider protection technology for the protection of e-scooters. © 2024 Elsevier Ltd