Automobile on-demand gear-shift control including driver characteristics

In this paper, a new automobile gear-shift control for Continuously Variable Transmissions (CVTs) is presented. Conventional gear-shift control is based on so-called maps. Its setting is generally conservative and drivers can change shift maps themselves by operating switches such as a gear-shift or tip switches. However, it is troublesome for the drivers who are not positive for driving to operate them. To overcome this problem, we design a new gear-shift control for CVTs considering the driver characteristics of gear-shift operation based on a Just-In-Time(JIT) modeling. In addition, we reduce the number of input variables of the JIT model by linear regression in view of the limits on memory capacity and computational load for on-line calculation. Furthermore, the output of JIT model is smoothed with the least squares B-spline interpolation. Our experiments show that it has downshifted automatically on demand without driver's gear-shift operation when approaching curves.