Multi-environment robust polarization navigation sensor and the parameter adaptive prediction method

As a type of photosensitive sensor, the polarization navigation sensor is susceptible to variations in external optical information. Experiments have demonstrated that light intensity is one of the factors influencing solution parameters. Considering that traditional polarization navigation sensor cannot detect light intensity effectively, a polarization sensor with light intensity detection capability is designed. Additionally, a neural network-based adaptive parameter method is proposed. By utilizing a neural network for the real-time solution degree of linear polarization (DLOP) and combining the light intensity information with polarization information from the new sensor, the method finally realizes the prediction of external solution parameters, thus enabling the solution model to adapt to the environmental variations. Finally, the effectiveness of the adaptive prediction method is verified by experiments. The results of outdoor experiments show that the adaptive prediction method reduces the mean square error (MSE) of angle of polarization (AOP) by approximately 42.11%, 44.78%, and 52.83% compared to the convention solution method under sunny, cloudy, and overcast conditions. •A novel polarization navigation sensor with the ability to detect light intensity is designed.•An adaptive predication method to enhance the robustness of sensors in multiple environments is proposed.•The MSE of the angle of polarization solution under overcast condition is 0.65°, with an improvement of 52.83%.