A Novel Acceleration Signal Processing Procedure for Cycling Safety Assessment

Area of Science:

• Urban planning and transportation safety
• Cyclist behavior analysis
• Data science and signal processing

Background:

• Urban cycling is hindered by safety concerns, necessitating identification of high-risk areas.
• Traditional crash data analysis is insufficient for cyclist safety due to data limitations.
• Modern technology offers new approaches for collecting cyclist behavior data.

Purpose of the Study:

• To analyze data requirements for identifying critical cycling safety events in urban environments.
• To evaluate the capability of bicycle-based data collection for cyclist safety.
• To determine the minimum GPS sampling frequency for detecting hard braking events.

Main Methods:

• Investigated three different GPS sampling frequencies to detect hard braking.
• Developed a novel signal processing procedure for speed and acceleration data.
• Applied wavelet transformation and Dynamic Time Warping (DTW) for noise reduction and signal alignment.

Main Results:

• A minimum GPS sampling frequency of 1Hz is necessary to accurately track rider behavior and detect safety events.
• The proposed signal processing, including DTW, effectively removes instrument noise and aligns signals.
• The Euclidean distance of DTW was used to optimize filter parameters.

Conclusions:

• Affordable hardware setups can effectively monitor cyclists' speed and acceleration.
• This research provides a foundation for developing proactive cyclist safety countermeasures.
• 1Hz GPS sampling is sufficient for capturing critical cyclist behavior data.