Fig. 1. (a) Schematic diagram of micro-LED backlight module; (b) schematic diagram of LED with DBR structure; (c) highly-reflective surface substrate; (d) etching structure of the receiver.

Fig. 2. Workflow for optimizing micro-LED backlight module: (a) the process of deep reinforcement learning and (b) the process of the virtual-realistic experiment.

Fig. 3. (a) Workflow of environment control agent and schematic diagram of the virtual-realistic experiment; (b) principle of kernel1: Gaussian and Lambertian reflection; (c) principle of kernel2: BSDF properties.

Fig. 4. Workflow of DDQN network.

Fig. 5. Virtual-realistic experiment: (a) single light pattern analysis; (b) module pattern analysis.

Fig. 6. Result of reinforcement learning: (a) uniformity for every iteration with reward function1; (b) uniformity for every iteration with reward function2; (c) uniformity for every iteration with reward function3; (d) the best result by reward function1; (e) the best result by reward function2; (f) the best result by reward function3.

Fig. 7. Demonstration of the micro-LED backlight module.

Fig. 8. Influence of DBR structure: (a) light pattern with state Sb3′, (b) light pattern with state Sb3, and (c) uniformity of Sb3 and Sb3′.

Fig. 9. Schematic of resonant loss for the micro-LED backlight module.

Table 1. Definition List of Action

Table 2. Definition List of State

Table 3. Definition List for Range of Parameters

Table 4. Results of Virtual-Realistic Experiment

Table 5. Best Uniformity for Different Reward Functions

Table 6. Work Efficiency of Designing Agent