Physically Interpretable Three-Phase Jump Planning and Experimental Validation for a 200 kg-Class Wheel-Legged Robot

Jumping capability provides wheel-legged robots (WLRs) with an effective means to traverse discontinuous terrains such as ditches and broken bridges. However, implementing jumping behavior on heavy-duty robots imposes higher requirements on motion planning in terms of computational efficiency, interpretability, and reliability. This article proposes a physically interpretable three-phase trajectory planning method tailored for heavy-duty WLRs. Based on a cart-mass model, the method constructs analytical trajectory expressions for the take-off, flight, and landing phases. The trajectories are driven by a small set of physically meaningful parameters, avoiding reliance on complex numerical optimization or learning-based strategies, and thus achieving an excellent balance between computational efficiency and deployment robustness. Comparative simulation studies with repeated trials and statistical reporting validate the effectiveness of the proposed complete three-phase framework. The proposed method shows clear advantages over a take-off-focused baseline in jump realization, take-off responsiveness, peak electro-hydraulic actuator force, and total joint energy consumption, while remaining competitive with an optimization-based three-phase planner, with a slightly lower but still close jump completion ratio together with a shorter take-off duration. Hardware experiments are conducted on a self-developed 200 kg-class wheel-legged robot, Rhino. The robot successfully performs repeated vertical jumps under no-load conditions, validating the deployment robustness of the proposed planning framework, and also demonstrates a run-and-jump exceeding 1.4 m while carrying a 100 kg payload, supporting its practical feasibility in a dynamic heavy-payload scenario.