This research studies a novel method of realizing a nonmechanical antilock braking system (ABS) controller for electric scooters (ESs) based on regenerative, kinetic, and short-circuit braking mechanisms. In which, a boundary layer speed control is proposed for a guarantee of the optimal slip ratio between tires and road surface. The antilock braking controller, combined with this controller, drives a low-side driving circuit to induce either an open-circuit or a short-circuit loop on the motor stator's coil to a load; it thus produces braking actions analogous to those in the conventional ABS control. The proposed ABS controller is practically realized. Improvement of the braking performance for the ABS action is further addressed via real-world experiments.Index Terms-Antilock braking system (ABS), boundary layer control, electric vehicle (EV), short circuit braking.

No abstract

This paper proposes a novel fast dehazing method based on dark channel prior. The dark channel prior is the phenomenon about normal outdoor images containing at least one low energy pixel around a block among three channels. It helps roughly estimating the air mediums thickness, which decreases the scene transmission. We propose a variation of optical model whose single atmospheric attenuation coefficient is replaced with three different ones corresponding to each channel. In the past, the big sliding window of dark channel prior for calculating the transmission map inevitably resulted in unacceptable computing burden and has to be narrowed down. However, hue offset will be resulted if the size of sliding window is reduced. Our proposed method eliminates the offset while greatly reducing the processing time. Then, a joint bilateral filter which replaces the original matting method is applied as our smooth measure to further refine the estimated transmission maps. Finally, we apply dehazing into driver assistance applications-lane and vehicle recognition. Robust lane recognition provides the information about if the input image to be fed into vehicle recognition should be conducted with dehazing to improve detecting rate.