pNMPC - A Code Generation Software Tool for Implementation of Derivative Free Parameterized NMPC Scheme for Embedded Control Systems

In this paper, we propose a derivative free pa-rameterized Nonlinear Model Predictive Control (pNMPC) code generation software (S/W) tool for embedded control systems. The proposed S/W tool serves two purposes a) Modeling and specification of the optimal control problem (OCP) and b) C code generation of a highly optimized, portable and efficient pNMPC controller suited for real time (RT) control of fast sampled systems. The underlying optimization module for the pNMPC controller is a SQP based black box optimization (BBO) solver which is specifically tailored to handle inequality constraints specified by the OCP. The S/W tool was completely programmed in C++ programming language and provides interface to MATLAB/Simulink environment with the help of MEX and C MEX-S function wrappers. Two benchmark problems were considered namely, Cart-pole stabilization problem and PVTOL probem and tests were conducted using the proposed S/W tool and also, compared against the ACADO toolkit to assess the performance, efficiency and evaluate the computation time and validate the proposed S/W in-spite of black box models. In total, the proposed S/W tool fares well and looks promising for practical RT embedded control