Abstract: Ratoon rice is a typical cultivation to allow for the growth of another crop of rice using dormant seedlings from the previous harvest. Seedling, raising, and transplanting of double-season rice can be removed to reduce the harm from pesticide residues. Cultivated land quality can also be protected at a low cost. Additionally, the quality and fullness of the second crop of ratoon rice are superior to that of double-season rice. The planting area of machine-harvested ratoon rice has increased rapidly in recent years. Therefore, most research has been focused on the harvest technology and equipment of ratoon rice. Particularly, ratoon rice harvesting is required for the low rolling and high stubble retention for optimal agronomic conditions. The rolling of rice stubble during harvest in the main season can significantly reduce the yield of ratoon rice in the ratoon season. Once the lodging occurs at the mature stage of crop growth, it is detrimental to the harvesting yield and quality. Therefore, the optimal path of harvesting is crucial to improve the efficiency of grain production for better harvesting with minimal loss. Concurrently, the volume limit of the grain bin in the harvesters and the location of the unloading points also exert a significant impact on the loss of harvesting and low rolling of the ratoon rice. Therefore, it is necessary to design the harvest path of the ratoon rice harvesters for the maximum yield in the main season. The optimal path for high-quality agricultural machinery is crucial to saving time and energy, particularly for better operation efficiency and quality with less soil rolling. In this study, a path planning was proposed, called the harvester grain bin capacitated arc routing (HGBCARP). Two modules consisted of information processing and path planning. The information processing module was used to convert the operation information (such as farmland boundaries, unloading point positions, crop lodging directions, and area positions) into a processable data form, and then transmit it to the path planning module. The optimal path planning was obtained after the tasks, such as the direction division of the operation line, the optimization of the operation line traversal sequence, the generation of turning modes, and the calculation of the rolling area. An improved genetic algorithm was utilized to evaluate the performance of three new types of ratoon rice harvesters, as well as two traditional rice harvesters. Three parameters of the field test were taken as evaluation indicators, including the length of the driving path, the rolling area, and the amount of harvested grain. A comparative simulation was then conducted under three field conditions using rotary and HGBCARP harvesting path planning. It was found that the rolling area of the HGBCARP harvesting path was 11.79% to 27.20% less than that of the rotary type. The rolling area was reduced to increase the yield of the main season rice by 1.64% to 1.95%. Additionally, the HGBCARP harvesting path was found to reduce the rolling area by 7.25% to 20.09%, compared with the rotary type. Field harvesting experiments were carried out using an electric unmanned crawler chassis on various harvesting paths. The HGBCARP harvesting path was found to reduce the rolling area by 11.21% to 28.03% and the path length by 6.81% to 23.46%, compared with the traditional cattle tillage reciprocating and rotary harvesting paths. The high effectiveness was achieved in the HGBCARP path planning. This finding can also provide a valuable reference for intelligent operation path planning.