Three-wheeled agricultural robots possess the advantages of high flexibility, strong maneuverability, and low cost. They can adapt to various complex terrains and operational environments, making them highly valuable in the fields of crop planting, harvesting, irrigation, and more. However, the horizontal stability of the three-wheeled agricultural robot chassis is compromised when working in harsh terrain, significantly impacting the overall operational quality and safety. To address this issue, this study designed a leveling system based on active suspension and proposed a stepwise leveling method based on an adaptive dual-loop composite control strategy (ADLCCS-SLM). Firstly, in the overall control of the three-wheeled chassis, a stepwise leveling method (SLM) was introduced. This method allows for rapid leveling by incrementally adjusting one or two suspensions, effectively avoiding the complex interactions between suspension components encountered in traditional methods involving the simultaneous linkage of three suspensions. Next, in terms of suspension actuator control, an adaptive dual-loop composite control strategy (ADLCCS) was proposed. This strategy employs a dual-loop composite control both internally and externally and utilizes an improved adaptive genetic algorithm to adjust critical control parameters. This adaptation optimizes the chassis leveling performance across various road conditions. Finally, the effectiveness of the proposed ADLCCS-SLM was validated through simulation and experimental testing. The test results showed that the control effect of the proposed method was significant. Compared to the traditional multi-suspension linkage leveling method based on PID, the peak values of pitch angle and roll angle were reduced by 31.8% and 33.3%, respectively.

In marine hot–humid and salt spray environments, shipborne snap-action limit switches operating under micro-current loads are prone to triggering failures caused by the accumulation of heterogeneous films on electrical contact interfaces, which can induce abnormal behavior in electromechanical systems. To address this issue, this study systematically investigates the failure mechanisms of micro-current limit switches using multimodal diagnostic approaches. The results demonstrate that the migration and accumulation of corrosion products and foreign contaminants within the microswitch unit promote the formation of high-resistance heterogeneous films at the electrical contact interfaces, severely impairing reliable electrical conduction. Electrical contact experiments further reveal that the contact behavior is strongly dependent on the current magnitude. When the current exceeds 2A, arc discharge generated during contact closure can effectively disrupt and remove the heterogeneous films, thereby restoring the electrical functionality of previously failed switches under subsequent micro-current operating conditions. Based on the identified failure mechanism, and inspired by the natural eye-cleaning behavior of crabs, a biomimetic press-and-wipe self-cleaning dual-redundant limit switch design is proposed. The design enables autonomous surface cleaning through controlled reciprocal wiping between the moving and stationary electrical contacts, effectively suppressing the formation and accumulation of high-resistance films at the source. Comparative salt spray and damp heat storage tests demonstrate that the proposed self-cleaning limit switch maintains stable and reliable electrical contact performance in simulated marine environments, significantly improving operational reliability and service life under micro-current loads. This work provides both mechanistic insights and a practical structural solution for enhancing the reliability of electrical contact components operating under low-current conditions in harsh marine environments.