With the advancement of touch-control technology and autonomous driving systems, handwriting input, as one of the key interaction methods for touchscreen devices, has been widely adopted in in-vehicle intelligent terminals. However, vehicle vibrations and sudden impacts in automotive environments often lead to more cursive strokes and missing components in handwriting input, significantly reducing the accuracy of handwritten text recognition. To address this issue, this study proposes a Chinese handwriting online recognition method with radical decomposition (CHORRD). This approach first employs cursive-style data augmentation to expand the original dataset, enhancing the base model robustness in recognizing cursive writing styles. Subsequently, the structure and radical sequence of the handwritten text are extracted, and distances are calculated based on the proposed head- and tail-confirmation ideographic description sequence (HTCIDS) to determine the complete character. Experimental results demonstrate that the proposed method effectively achieves radical decomposition and character recognition in simulated vehicle vibration environments, significantly improving online handwriting recognition performance, with a recognition rate exceeding 90%.