Fast Model Predictive Control Using Online Optimization

摘要: A widely recognized shortcoming of model predictive control (MPC) is that it can usually only be used in applications with slow dynamics, where the sample time measured seconds or minutes. well-known technique for implementing fast MPC to compute entire law offline, which case online controller implemented as a lookup table. This method works well systems small state and input dimensions (say, no more than five), few constraints, short horizons. In this paper, we describe collection methods improving speed MPC, using optimization. These custom methods, exploit particular structure problem, action on order 100 times faster uses generic optimizer. As an example, our computes actions problem 12 states, 3 controls, horizon 30 steps (which entails solving quadratic program 450 variables 1284 constraints) around 5 ms, allowing carried out at 200 Hz.