<p>This study proposes a novel 6-degree-of-freedom leg mechanism with a series–parallel mechanism for an unmanned deformable vehicle. This mechanism not only overcomes the challenges of high moment of inertia and insufficient load-bearing capacity of the leg mechanism of bipedal robots, but also can deform to meet the requirements of folding and unfolding movements. The kinematic model and the Lagrangian dynamic model for leg mechanism motion were established after designing the leg mechanism. The reconfiguration from the vehicle state to the support state was analyzed for motion interference by solving inverse kinematic models. Additionally, the joint motion planning during the reconfiguration from the support state to the humanoid state and the gait planning during walking were designed. In order to improve the motion stability and reduce energy consumption, ant colony optimization algorithm was used to optimize the thigh length and calf length. On this basis, the Non-linear Programming by Quadratic Lagrangian (NLPQL) optimization algorithm was used to optimize the wall thickness of the thigh and calf. Finally, experimental results show that the leg mechanism meets the design requirements for motion, ensures motion stability, and reduces motion energy consumption.</p>