Answers are grounded in this paper and cite their sources.
Demonstrates effective gear selection in bicycles, suggesting a cost-effective alternative to power meter-based systems.
High-end bicycles are increasingly equipped with electronic drivetrains that shift automatically. Current commercial systems infer the “right” gear from wheel speed, cadence and rider power input, with the latter obtained from an expensive crank- or pedal-based power meter. The price and integration effort of power meters limit the spread of automatic shifting beyond top-tier models. This paper proposes inclination-based automatic shifting, a concept that eliminates the power meter altogether. A single, low-cost inertial-measurement unit mounted on the frame provides real-time pitch data from which the road gradient is estimated; combined with speed and cadence, the gradient captures most of the situational context that a power-meter offers. A test bicycle was equipped with both a commercial power meter and an off-the-shelf inertial measurement unit and collected three hours of ride data over varied terrain, which included the rider-selected gear. Pitch, speed, and cadence were subsequently utilized to train state-of-the-art time-series classification algorithms, including Rocket, MiniRocket, CNN, and LSTM models, to predict the target gear. These predictions were then compared to those derived from a combination of power meter torque, speed, and cadence-based predictions. The findings suggest that inclination sensing offers a viable and cost-effective approach to mass-market automatic shifting, opening new design opportunities for bicycle manufacturers. Additionally, it presents a data-driven, indirect method for user context estimation, serving as a valuable alternative to consider in the design of new smart products.
No takes yet. Share an insight, caveat, or question.
Oberle et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: