Abstract:Traditional road mapping relies on manual operations, which are inefficient and error-prone, necessitating the automated tools. An AI code generation model based on Natural Language Processing (NLP) can realize the intelligent processing of CAD files in road mapping by using AutoLISP language. Taking the terrain map processing in road mapping as an example, a CAD terrain processing program based on LISP language is developed to achieve the functions such as contour line style optimization and layer classification. By generating LISP script logic through artificial intelligence and integrating road engineering drafting standards, an automated processing framework is constructed. The program significantly reduces the terrain map processing time, lowers the manual error rates, and demonstrates the practicality through the validation of road projects in Shenzhen. The rigid constraints of traditional script programming is broken through, and the intelligent transformation of natural language instructions into professional design logic is realized, which provides an innovative technical pathway for digital delivery in road engineering.