<p>Accurate estimation of line-of-sight (LOS) angular velocity is essential for successful interception during the terminal homing phase of guided missile systems. However, measurements from onboard seekers are often corrupted by noise, and the dynamic behavior of high-speed targets is both nonlinear and non-Gaussian. To ensure robust and stable state estimation under these conditions, this paper proposes a Huber-based extended Kalman filter (HEKF), which enhances robustness to outliers. Although this approach improves estimation reliability, the iterative optimization required to minimize the Huber cost function increases computational complexity. To overcome the computational bottleneck, the HEKF is implemented on a field-programmable gate array (FPGA). Specifically, the Huber function is implemented as a custom intellectual property (IP) to enable high-throughput, low-latency computation. Evaluation results confirm that the FPGA implementation accelerates the overall HEKF by 3.71 × and the Huber IP by 6.79 × over the processing system (PS). While maintaining comparable denoising performance and estimation accuracy, the proposed method significantly improves computational efficiency through FPGA-based parallelization.</p>