Abstarct: The Spring-Loaded Inverted Pendulum (SLIP) model limits the controllability and mobility of the robot in bipedal walking due to the omission of the ankle joint. This research proposes a new walking pattern tixtled "Optimal Design of Biped Robot Walking baxsed on an Improved Inverted Pendulum Model," also known as the Variable Spring-Loaded Inverted Pendulum with Finite Foot (VSLIP-FF) model for biped robots. By adding a finite-sized foot and a one-degree-of-freedom ankle joint for each leg and making the leg stiffness adjustable, the VSLIP-FF model can be used to achieve flexible bipedal walking in complex environments. Inspired by human walking characteristics, an adaptive leg extension and contraction strategy is proposed for gait planning to mimic the role of the ankle joint. Research on bipedal walking with the VSLIP-FF model on flat and inclined surfaces can improve our understanding of stability and balance control in robots. Investigating how robots adapt to sloped surfaces could lead to the development of more advanced control algorithms that maintain stability and balance even with sudden changes in the walking surface. Additionally, we will examine the factors affecting speed and proper walking for the VSLIP-FF model on flat and inclined surfaces. Furthermore, research on walking robots with the VSLIP-FF model on flat surfaces, considering static and sliding friction forces, can enhance our understanding of stability and balance control in robots. For a given walking pattern, the necessary friction coefficient can be calculated to allow the robot to perform the expected movement and bring the model closer to the real human walking behavior at each step. This stability and adaptability are crucial for the practical use of bipedal robots in various terrains. Moreover, to ensure walking stability, the Foot Rotation Indicator (FRI) point is used as a stability criterion to prevent the robot from falling. The Particle Swarm Optimization (PSO) algorithm is used to optimize the desired paths of the Center of Mass (CoM) and feet in a complete walking cycle. The results of the simulation of the model in the mentioned conditions are that the model was able to maintain its stability and behaves like a human being and takes a longer step length in less time than the SLIP model. Simulation results under the mentioned conditions show that the model maintained its stability on both flat and inclined surfaces with a slope of 0.08, displaying human-like behavior and covering longer steps in less time compared to the SLIP model. Additionally, friction conditions for the model were considered on flat surfaces, and baxsed on the model's parameters, if the coefficient of friction is greater than or equal to 0.04725, the friction between the robot’s foot contact surface and the ground is static friction; otherwise, it is dynamic friction.