A two-stage learning network for PVINS modeling and fusion estimation in challenging environments

Key Points

• The two-stage learning network significantly improved fusion estimation precision in challenging environments, indicating better model performance.
• Achieving up to 95% accuracy in complex conditions illustrates the network's robustness and reliability.
• Assessment using a two-stage learning network highlighted its effectiveness in PVINS modeling under various environmental factors.
• These findings suggest that the two-stage approach might facilitate better applications in real-world challenging situations.