Multi-source data fusion SLAM systems and robot path planning have emerged as prominent research focuses in the agricultural automation domain. Low-cost and semantically enriched SLAM systems have demonstrated significant practical value in production environments. Over the past decades, although remarkable progress has been made in both multi-source data fusion and SLAM technologies, a series of unresolved challenges persist. This literature review aims to systematically delineate the evolutionary trajectories and current advancements in these two fields, thereby establishing a robust theoretical foundation and technical reference for subsequent research. The review will separately analyze the critical issues and challenges confronting multi-source data fusion SLAM systems and robot path planning in practical applications. Key concerns include: Trade-offs between system accuracy and real-time performance, particularly in resource-constrained agricultural settings. Data alignment and calibration in dynamic environments, where heterogeneous sensor inputs may degrade system robustness due to temporal-spatial misalignments or environmental perturbations. For each challenge, innovative methodologies proposed by researchers will be critically examined. Examples include adaptive sensor fusion frameworks, uncertainty-aware optimization algorithms, and dynamic environment modeling techniques. Finally, the review concludes with a forward-looking perspective on multi-source data fusion SLAM systems and path planning. Future research directions are anticipated to prioritize lightweight architectures for edge deployment and collaborative multi-robot systems to enhance scalability and efficiency in large-scale agricultural operations. These advancements are expected to bridge existing gaps between theoretical innovations and industrial applications, ultimately driving the next generation of intelligent agricultural robotics.