297
Views
0
CrossRef citations to date
0
Altmetric
Research Article

Deep Recurrent Reinforcement Learning for Intercept Guidance Law under Partial Observability

, , &
Article: 2355023 | Received 15 Jul 2023, Accepted 30 Apr 2024, Published online: 16 May 2024

Figures & data

Figure 1. The proposed guidance framework based on deep recurrent reinforcement learning.

Figure 1. The proposed guidance framework based on deep recurrent reinforcement learning.

Figure 2. The workflow of the TD3 algorithm.

Figure 2. The workflow of the TD3 algorithm.

Figure 3. Deep Recurrent Reinforcement Learning.

Figure 3. Deep Recurrent Reinforcement Learning.

Figure 4. Engagement geometry.

Figure 4. Engagement geometry.

Table 1. Training procedure of the proposed DRRL guidance law.

Figure 5. Ornstein–Uhlenbeck noise (μ=0,θ=0.15,σ=0.2).

Figure 5. Ornstein–Uhlenbeck noise (μ=0,θ=0.15,σ=0.2).

Table 2. Initial parameters for engagement.

Figure 6. Learning curves.

Figure 6. Learning curves.

Table 3. Test results in the training environment.

Figure 7. Interception trajectories of the APN, SMC and DRRL methods.

Figure 7. Interception trajectories of the APN, SMC and DRRL methods.

Figure 8. Performance comparison of APN, SMC and DRRL in test scenarios.

Figure 8. Performance comparison of APN, SMC and DRRL in test scenarios.

Figure 9. Comparison of learning curves when the model is trained with different values of m.

Figure 9. Comparison of learning curves when the model is trained with different values of m.

Table 4. Performance comparison with different lengths of observation sequence input.

Data Availability Statement

The authors confirm that the data supporting the findings of this study are available within the article.