Semi-automated transmission control for motorcycle gearshift: design, data-driven tuning and experimental validation

TL;DR

This paper tackles gearshift control for Semi-Automated Manual Transmission (S-AMT) in two-wheeled vehicles by combining an open-loop engine torque and clutch-by-wire (CBW) control with a data-driven calibration workflow. It introduces a Constrained Bayesian Optimization framework to tune a small set of key parameters and validates the approach on a Ducati Diavel prototype, demonstrating improved gearshift smoothness and repeatability over a baseline Quick-Shift strategy. Key contributions include a detailed multi-domain gearshift model, objective and constraint definitions for tuning, an experimental assessment of convergence and parameter sensitivity, and a demonstration that CBO outperforms Random Search in acquiring high-quality parameter sets. The proposed method is generalizable to other shifts and can be deployed in industrial settings, offering fast, real-time capable tuning with post-shift calibration that does not require full powertrain redesign.

Abstract

This brief addresses the gearshifting problem for Semi-Automated Manual Transmissions (S-AMT) in powered two-wheelers, a powertrain setup that allows fast and smooth gear shifts with minimal modifications to the traditional manual powertrain layout. We show that with a proper synchronization between the electronic clutch and engine torque, excellent gearshift performance can be obtained, but requires precise parameters calibration. We thus propose the use of a data-driven approach with Constrained Bayesian Optimization to optimize control parameters. The procedure's effectiveness is demonstrated on a real vehicle, assessing performance in terms of optimality, convergence rate, and repeatability.

Figures (9)

• Figure 1: Pictorial representation of the electro-hydraulic clutch-by-wire (CBW) system.
• Figure 2: Control strategy and data-driven parameter optimization framework representation.
• Figure 3: Time domain response of a QS-based logic (black) compared with respect to the optimized mixed QS-CBW (red) solution.
• Figure 4: Measured and estimated cost function values during an automatic optimization campaign.
• Figure 5: Samples distribution over the three-dimensional search space $\Theta$, along with the optimal candidate and the best visited query. Shaded areas indicate infeasible regions that do not meet constraints. Each plot displays the colormap of the estimated cost function, varying two parameters, with the third held at its optimal value.