Bionic Energy-Efficient Inverse Kinematics Method Based on Neural Networks for the Legs of Hydraulic Legged Robots

Hydraulic legged robots, with advantages such as high load capacity and power density, have become a strategic driving force in advancing intelligent mobile platform technologies. However, their high energy consumption significantly limits long-duration endurance and efficient operational performance. In this paper, inspired by the excellent autonomous energy-efficient consciousness of mammals endowed by natural evolution, a bionic energy-efficient inverse kinematics method based on neural networks (EIKNN) is proposed for the energy-efficient motion planning of hydraulic legged robots with redundant degrees of freedom (RDOFs). Firstly, the dynamic programming (DP) algorithm is used to solve the optimal joint configuration with minimum energy loss as the goal, and the training data set is generated. Subsequently, the inverse kinematic model of the leg with minimum energy loss is learned based on neural network (NN) simulation of the autonomous energy-efficient consciousness endowed to mammals by natural evolution. Finally, extensive comparative experiments validate the effectiveness and superiority of the proposed method. This method not only significantly reduces energy dissipation in hydraulic legged robots but also lays a crucial foundation for advancing hydraulic legged robot technology toward high efficiency, environmental sustainability, and long-term developmental viability.