Traditional approach for coordinate transformation is a level-wised process where several sets of equations obtained by geographical survey are invoked, which makes the whole process complicated and introduces the accumulation of calculating errors. This paper presents a genetic-based approach for finding symbolic formulas for transforming coordinate signals received from GPS. In this way, the receiver's position can be obtained much faster with less computational complexity. We test this concept on different areas in Taiwan and find several new equations that can effectively and better produce accurate coordinates