A Quadruped Robot Motion Controller Based on Adaptive Nonlinear MPC
Quadruped robots have the advantages of flexibility and large loads,and can be applied in complex environments such as geological exploration and disaster relief.Nonlinear MPC has recently shown superior performance in this field,but requires highly accurate models to achieve maximum performance.Uncertainties such as unmodeled loads and parameter mismatches can reduce system performance.This paper proposes a L1 adaptive control approach to address the problem of parameter mismatch in nonlinear MPC of quadruped robots.By identifying model uncertainties online and compensating for them,the performance of nonlinear MPC is significantly improved.Finally,through experimental verification,this method can effectively compensate for various model uncertainties,and significantly reduce the motion tracking error of quadruped robots under large unknown disturbances without any gain adjustment.
万方数据
维普
