Camera–magnetic fusion for lateral coil alignment in dynamic wireless charging on roads with degraded lane markings

Dynamic wireless charging of electric vehicles loses coupling rapidly as the receiver drifts sideways over the road pads, so the transmitter needs a reliable estimate of lateral offset. Camera methods depend on visible road markings, which are often worn, interrupted or absent on Indian roads, while magnetic methods work only over an energised pad. This paper develops and evaluates, in simulation, an estimator that combines both. A forward camera reads the lane centre, a two-axis coil at the receiver measures the in-plane angle of the pad field, and a gyro bridges the gaps, all within a four-state extended Kalman filter. The field angle is shown to vary monotonically with offset over the full 70 cm range examined and to be insensitive to transmitter current. The filter also estimates the unknown offset between painted lane and buried pads, which is unobservable from the camera alone. The estimate drives feedforward inverter control with a slow feedback trim. Camera measurements come from rendered road images processed by a real lane detector under five conditions, from fresh markings to near-unmarked surfaces, over 150 runs at 30 to 90 km per hour. The fused estimator reached 2.8 cm root-mean-square lateral error overall and 0.9 cm over powered sections, against 5.8 cm for magnetic sensing alone and 17.0 cm for the camera alone. Power-tracking error fell from 12.0 percent without sensing to 4.6 percent. Fusion gave no benefit on unmarked roads, outlier gating caused occasional lock-out on patchy roads, and the sensitivity analysis attributes most of the remaining power error to the unknown air gap. Hardware validation and air-gap estimation are the next steps.