Figure 5
The diagram presents a workflow connecting a “Tree Growth Game” environment with a “D R L Model” for evaluating tree maintenance strategies. The diagram is divided into two large colored regions. The left pink region is labeled “Tree Growth Game” and the right blue-gray region is labeled “D R L Model”. On the left side, a rounded rectangular panel titled “Environment” contains a three-dimensional visualization in a three-dimensional coordinate plane labeled “Tree Growth Simulation”. A rightward arrow points to a curved panel labeled “Future State of The Tree” under “D R L Model”. Inside the curved panel labeled “Future State of The Tree” is a green voxel-style three-dimensional tree structure. Purple text below the panel labels it as “State S subscript t plus 1”. To the lower-right side of this panel is another curved white panel labeled “Targeted State of the Tree”. Inside this panel is a flattened green voxel-style tree canopy structure. The purple text below labels it as “Goal S subscript T A R”. Between the two state panels is a circular comparison symbol containing an “X”. The text above the comparison node reads “R S subscript t plus 1 comma S subscript T A R”. A rightward arrow from “Future State of The Tree” and an upward arrow from “Targeted State of the Tree” connect to this node. Two dashed horizontal arrows extend rightward from the comparison node toward a vertical divider line. The upper dashed line is labeled “Reward R subscript t plus 1”. The lower dashed line is labeled “State S subscript t plus 1 minus S subscript T A R”. To the right of the divider, their corresponding arrows labeled “R subscript t” and “S subscript t minus S subscript T A R” point right to another panel labeled “Evaluation”. On the far right is the rounded rectangular panel labeled “Evaluation”. Inside the panel is a neural network-style diagram composed of connected circular nodes. The text below the panel labels it as “Agent”. A feedback arrow extends upward from the “Evaluation” panel and points toward a white box near the upper-left area labeled “Decision in Strategies for Tree Maintenance”. Above the feedback arrow is the purple expression “Q S subscript t comma A subscript t”. Purple text below the decision box reads “Action A subscript t”. This arrow continues leftward towards the “Tree Growth Simulation” under “Environment”.Structure of the DRL model in decision-making