Quantifying the Uncertainty in Bicycle-Computer Speed Measurements

Bicycle computers record and store kinematic and physiologic data that can be useful for forensic investigations of crashes. The utility of speed data from bicycle computers depends on the accurate synchronization of the speed data with either the recorded time or position, and the accuracy of the reported speed. The primary goals of this study were to quantify the temporal asynchrony and the error amplitudes in speed measurements recorded by a common bicycle computer over a wide area and over a long period. We acquired 96 hours of data at 1-second intervals simultaneously from three Garmin Edge 530 computers mounted to the same bicycle during road cycling in rural and urban environments. Each computer recorded speed data using a different method: two units were paired to two different external speed sensors and a third unit was not paired to any remote sensors and calculated its speed based on GPS data. We synchronized the units based on the speed signals and used one of the paired speed sensors as a reference. We found that the time, position, and speed recorded in the data files were not synchronized, although the lag between the speed and position data was consistently within 0 to 3 s. We then generated probability distributions that quantified the bias (median) and uncertainty (95th percentile interval) in the internal and external measures of speed. The biases were -0.10 m/s for the internally calculated speed and -0.02 m/s for the externally calculated speed. The uncertainty ranged from 1.14 m/s below to 0.47 m/s above the reference speed for the internally calculated speed, and from 0.51 m/s below to 0.47 m/s above the reference speed for the externally measured speed. This study provides useful baseline data for quantifying the temporal asynchrony, bias, and uncertainty of speed measurements recorded by bicycle computers.