This paper presents a novel approach for tire load identification in Automated People Mover (APM) vehicles using a Multi-Sensor KalmanNet Fusion (MSKNF) framework. Traditional methods of tire load estimation, which rely on either purely physical models or data-driven approaches, face challenges in real-world conditions due to uncertainties in parameters like temperature, grip, and system dynamics. The MSKNF framework combines the strengths of multi-modal sensor data and data-driven techniques, utilizing an Interactive Multi-Model (IMM) approach to improve estimation accuracy and robustness. The framework processes sensor data from displacement and acceleration measurements and uses KalmanNet to learn Kalman gains, thereby mitigating noise and enhancing the reliability of the estimates. This study demonstrates the efficacy of the proposed approach using a Bombardier INNOVIA 300-type APM vehicle, showing significant improvements over traditional Extended Kalman Filter (EKF) models in highly nonlinear environments. The fusion of multi-modal sensor data provides a comprehensive and reliable estimation method for ensuring vehicle safety and performance.