Ground-penetrating radar (GPR) is often used to locate utilities like pipes and cables in the subsurface. In many GPR surveys, a full grid pattern is performed even if the goal is to locate just a few utilities. This approach results in extended operational time. The aim of this project was to develop an adaptive search pattern for utility mapping that uses as short a path as possible, and that could be applied to an autonomous single-channel GPR system in the future. The search pattern was developed and simulated with Python. Animation and static plots were generated to demonstrate and evaluate its performance. A few pipe configurations were created for testing and performance evaluation. The result of the project is a search pattern that can traverse and map a network of utilities. It is able to find and map most pipes in test configurations but there are corner cases that need more attention. It can be used as a guide for a human operator, which means that a computer tells the operator how to move the system. The efficiency of the search pattern was evaluated. On a fairly sparse network of pipes, it can reduce the time consumption by at least 50% compared to the corresponding grid search.