AI-Assisted Selective Harvesting and Smart Forest Management: A State-of-the-Art Review of Multimodal Sensing and Decision-Support Approaches

Selective harvesting requires tree-level decisions that balance operational productivity, stand development, and sustainable forest management, yet current AI and sensing studies often remain fragmented across remote sensing, machine perception, optimization, and forestry domains. This review synthesizes research on AI-assisted selective harvesting and smart forest management, with emphasis on multimodal sensing, decision support, and harvester-oriented deployment. A PRISMA-informed scoping review was conducted using Scopus, Web of Science Core Collection, IEEE Xplore, ScienceDirect, and SpringerLink, resulting in 82 studies retained for qualitative synthesis. The reviewed literature was organized into eight thematic groups covering smart forestry, harvesting operations, RGB-based perception, LiDAR and point-cloud processing, multimodal fusion, edge deployment, decision-support systems, and forecasting-oriented digital forestry. The analysis shows that RGB imaging provides semantic tree recognition, LiDAR enables spatial localization and structural assessment, and decision-support methods can translate tree-level observations into transparent cut/keep recommendations. However, integrated harvester-mounted systems remain underdeveloped, particularly regarding real-time RGB–LiDAR fusion, operator-facing recommendations, and forecasting integration. This review proposes a reference architecture for human-in-the-loop AI-assisted selective harvesting and identifies future research priorities for field validation and smart forest-management integration.