Muiti-objective Gait Optimizntion of Lower-limb Exoskeleton Robot

More and more lower-limb exoskeleton robots are used in daily life, so some practical problems must be considered, such as gait optimization which can make robots more human-like and more effcient. However, t here are few stu dies on gai t opti miza ti on. This paper proposes a general multi-objective gait optimization framework for the lower-limb exoskeleton robot. Firstly, an individualized gait pattern generation (IGPG) model is used to generate natural gait data as referential data. Secondly, the human-like cost func tion is cons true ted by a similarity measure in time domain of gait data generated by a fast parameterized gait planning method (FPGPM) and referential natural data, while the energy cos t function is cons true ed by calcula ting the total elec trical energy used to achieve gai t. Then, the weigh ted sum of human-like cost function and energy cost function is taken as the total cost function. Lastly, the parameters of FPGPM are optimized by an evolutionary algorithm based on parallel computing. The resuIts are verified by a lightweight lower-limb exoskeleton robot (LLEX), and experimen tai resu Its demons trate that t his approach can increase the energy efficiency of LLEX by 34.25% and the similarity with natural gait by 80.10% when compared with non-optimized gait.