Toward Dependable Embedded Model Predictive Control

Abstract

While model predictive control (MPC) is the industrially preferred method for advanced control in the process industries, it has not found much use in consumer products and safety-critical embedded systems applications in industries such as automotive, aerospace, medical, and robotics. The main barriers are implementability and dependability, where important factors are implementation of advanced numerical optimization algorithms on resource-limited embedded computing platforms and the associated complexity of verification. This challenge comes from a requirement of the use of ultrareliable hardware and software architectures in safety-critical applications, low-cost hardware in consumer products, or both. This paper surveys the state-of-the-art in the emerging field of dependable embedded MPC, and discusses some key challenges related to its design, implementation, and verification. A novel result is the study of a simulator-based performance monitoring and control selection method that monitors and predicts MPC performance and switches to a highly reliable backup controller in cases when the MPC experiences performance issues.