Wheel loaders play a crucial role in daily production and transportation. With the rapid development of intelligence in passenger vehicles, freeing loader operators from high-risk and repetitive tasks has become a pressing issue. This paper presents a novel and efficient path planning and tracking framework tailored to the unique body structure and specific operating environment of loaders. We improve the Hybrid A* search algorithm based on the operational characteristics of loaders and integrate it with dynamically updated grid maps to enable the autonomous planning of loader operating paths in unstructured environments, meeting the efficiency requirements of production. Additionally, to address the challenge of poor trajectory tracking control accuracy caused by hydraulic articulated steering, we propose a new loader trajectory tracking controller based on the idea of hierarchical control. We use an extended state observer to compensate for unknown disturbances in the steering execution layer and employ fuzzy fractional-order PID to handle the nonlinearity of loaders. Field experiments using the proposed approach demonstrate that loaders can autonomously and in real-time complete tasks in dynamically changing operating scenarios.

To address the challenges arising from external load uncertainties and the limitations of hydraulic system modeling accuracy, this article proposes a novel cascaded nonlinear integral sliding mode control (CNISMC) scheme, enhanced by an improved extended state observer (ESO). The proposed solution, in contrast to conventional sliding mode control (SMC) approaches that are susceptible to chattering, incorporates a novel convergence law and a modified sliding surface, thereby substantially mitigating chattering effects and enhancing transient performance. Furthermore, by implementing a cascaded control structure, the hydraulic system is segmented into two subsystems, each aligned with specific disturbance sources and integrated with an advanced ESO for accurate disturbance compensation. The proposed scheme is validated through a series of simulations and experimental studies, showing a significant improvement in control performance compared to conventional controllers.

In response to the challenge of directly measuring excavation resistance, this study introduces a predictive method based on a parallel hybrid deep neural network. Initially, addressing the difficulty in obtaining excavation resistance data, a joint electromechanical-hydraulic simulation model for loaders was developed to generate a dataset for excavation resistance. Considering the presence of local spatial features and long-term dependencies in the sequence, a residual structure was incorporated into the Convolutional Neural Network (CNN) to enhance the focus on significant features within the resistance data during spatial feature extraction. Additionally, Gated Recurrent Units (GRUs) were employed to capture temporal features. Subsequently, these two types of features were separately unfolded and then integrated through a fully connected layer to facilitate feature fusion, thereby avoiding the degradation of original data features typical in traditional serial architectures. Finally, the predictive performance of the proposed method was evaluated in comparative experiments with CNNs, GRUs, Transformers, and serial-structure hybrid neural networks. The results demonstrated an improvement in prediction accuracy compared to other methods. Specifically, the WMAPE values for longitudinal resistance (Fx), vertical resistance (Fz), and resistance torque (My) were 2.89%, 3.5%, and 5.78%, respectively. The proposed method decreases the WMAPE by 45.47%, 44.35%, and 26.8% in comparison to the serial architecture. The mean R² values were 0.982, 0.977, and 0.962, respectively.