基于遗传算法的两足步行机器人步态优化

将具有２４ 个自由度的机器人ＪＦＨＲ简化为７ 连杆机构，建立了描述机器人姿态的位置向量．用三次多项式拟合机器人髋关节和踝关节的位置轨迹，并通过动力学模型建立能量消耗表达式，得到能量最优的步态，即多项式系数的获得就可以表示为一个多变量最小值的优化问题，最后应用遗传算法（Ｇｅｎｅｔｉｃ Ａｌｇｏｒｉｔｈｍ ， ＧＡ）获得最优解．在ＧＡ设计中，将每一个需要优化的参数用１０位二进制数表示，种群中染色体的个数为５０，演化的代数固定为１００，杂交率和变异率分别定为０．８和０．０４．对平地步行和斜坡步行进行了仿真．

Gait Optimization of Biped Robot Using Genetic Algorithm

A 24 DOF humanoid robot (JFHR) was chosen for gait optimization. The dynamic model of JFHR was established as a seven link biped. Motion of the hip and feet during a regular step was then modeled by the 3rd order polynomials, and the energy consumption was obtained to form the robot dynamic equation. Therefore, the energy optimal gait, which is represented by the coefficients of polynomials, may be expressed as a multi variable minimization problem, and genetic algorithm was applied to obtain the optimal solution. In the genetic algorithm designing, every parameter to be optimized is encoded as 10 bit binary string, the total length of each chromosome is decided by the parameter number. The population for each generation is 50, the evolvement generation is fixed to 100. The crossover rate and mutation rate are set to 0.8 and 0.04 respectively. The simulation of walking on level ground and incline was presented.