Hierarchical steering coordination for reconfigurable modular wheeled platforms via virtual axle motion representation.

Abstract

Reconfigurable modular wheeled platforms (R-MWPs) enable flexible axle configurations and steering behaviors, making them promising for heavy-duty logistics and warehousing. However, architectural reconfiguration and modular heterogeneity fundamentally challenge accurate and synchronized execution in multi-axle systems. To address this problem, this paper proposes a hierarchical steering coordination framework based on a virtual-axle motion representation. At the platform level, a twist-based interface reformulates discrete steering-mode switching as continuous instantaneous center of rotation (ICR) evolution, enabling feasibility-aware motion iteration under physical angle and rate constraints. At the module level, a generalized topology is derived to characterize both intra- and inter-axle coupling, based on which a predefined-time consensus controller is developed to guarantee synchronized tracking under structural and environmental heterogeneity. At the actuator level, a gated deep deterministic policy gradient (DDPG) compensator is integrated to mitigate structural nonlinearities while preserving closed-loop stability and interpretability. The proposed framework is validated through real-vehicle experiments on 2-, 4-, and 6-wheel platforms, demonstrating that the scheme achieves precise execution, continuous transitions, and scalable coordination.