Real-Time Bio-Inspired Polarization Heading Resolution System Based on ZYNQ Heterogeneous Computing

Polarization navigation is an emerging navigation technology, that exhibits significant advantages, including strong anti-interference capability and non-cumulative errors over time, making it highly promising for applications in aerospace, autonomous driving, and robotics. To address the requirements of high integration and low power consumption for tri-directional polarization navigation sensors, this study proposes a system-on-chip (SoC) design solution. The system employs the ZYNQ MPSoC (Xilinx Inc., San Jose, CA, USA) as its core, leveraging hardware acceleration on the Programmable Logic (PL) side for three-angle polarization image data acquisition, image preprocessing, and edge detection. Simultaneously, the Processing System (PS) side orchestrates task coordination, performs polarization angle resolution, and extracts the solar meridian via Hough transform. Experimental results demonstrate that the system achieves an average heading angle output time interval of 9.43 milliseconds (ms) with a mean error of 0.50°, fulfilling the real-time processing demands of mobile devices.