Efficient representation and approximation of model predictive control laws via deep learning

Abstract

We show that artificial neural networks with rectifier units as activation functions can exactly represent the piecewise affine function that results from the formulation of model predictive control of linear time-invariant systems. The choice of deep neural networks is particularly interesting as they can represent exponentially many more affine regions compared to networks with only one hidden layer. We provide theoretical bounds on the minimum number of hidden layers and neurons per layer that a neural network should have to exactly represent a given model predictive control law. The proposed approach has a strong potential as an approximation method of predictive control laws, leading to better approximation quality and significantly smaller memory requirements than previous approaches, as we illustrate via simulation examples. We also suggest different alternatives to correct or quantify the approximation error. Since the online evaluation of neural networks is extremely simple, the approximated controllers can be deployed on low-power embedded devices with small storage capacity, enabling the implementation of advanced decision-making strategies for complex cyber-physical systems with limited computing capabilities.

arXiv comment13 pages, 7 figures

References in corpus (8)

• Adam: A Method for Stochastic Optimization
• On the Number of Linear Regions of Deep Neural Networks
• Highway Networks
• Learning an Approximate Model Predictive Controller with Guarantees
• Embedded Online Optimization for Model Predictive Control at Megahertz Rates
• Depth-Width Tradeoffs in Approximating Natural Functions with Neural Networks
• Bounding and Counting Linear Regions of Deep Neural Networks
• Computing control invariant sets is easy

Cited by in corpus (37)

• Safe and Fast Tracking on a Robot Manipulator: Robust MPC and Neural Network Control
• Deep Learning-Based Model Predictive Control for Resonant Power Converters
• Probabilistic performance validation of deep learning-based robust NMPC controllers
• Reliably-stabilizing piecewise-affine neural network controllers
• DeepOPF: A Deep Neural Network Approach for Security-Constrained DC Optimal Power Flow
• Improved uncertainty quantification for neural networks with Bayesian last layer
• Neural Lyapunov Differentiable Predictive Control
• Learning Lyapunov Functions for Piecewise Affine Systems with Neural Network Controllers
• Parallel Explicit Model Predictive Control
• Optimal Navigation in Microfluidics via the Optimization of a Discrete Loss
• Imitation Learning with Stability and Safety Guarantees
• Near-Optimal Rapid MPC using Neural Networks: A Primal-Dual Policy Learning Framework
• Approximate non-linear model predictive control with safety-augmented neural networks
• Error bounds for maxout neural network approximations of model predictive control
• Safe and Near-Optimal Policy Learning for Model Predictive Control using Primal-Dual Neural Networks
• Closed-loop training of static output feedback neural network controllers for large systems: A distillation case study
• Imitation Learning of MPC with Neural Networks: Error Guarantees and Sparsification
• DeepOPF: Deep Neural Network for DC Optimal Power Flow
• Neural Network Based Model Predictive Control for an Autonomous Vehicle
• Tailored neural networks for learning optimal value functions in MPC
• Factor Graph-Based Planning as Inference for Autonomous Vehicle Racing
• FedCoSR: Personalized Federated Learning with Contrastive Shareable Representations for Label Heterogeneity in Non-IID Data
• Learning optimal controllers: a dynamical motion primitive approach
• Tailored max-out networks for learning convex PWQ functions
• Polytopic Input Constraints in Learning-Based Optimal Control Using Neural Networks
• A Sensitivity-based Data Augmentation Framework for Model Predictive Control Policy Approximation
• Deep-MPC: A DAGGER-Driven Imitation Learning Strategy for Optimal Constrained Battery Charging
• Some manifold learning considerations towards explicit model predictive control
• AReN: Assured ReLU NN Architecture for Model Predictive Control of LTI Systems
• In Proximity of ReLU DNN, PWA Function, and Explicit MPC
• Error-free approximation of explicit linear MPC through lattice piecewise affine expression
• Bridging the Sim-to-real Gap: A Control Framework for Imitation Learning of Model Predictive Control
• Data-Driven Predictive Control Towards Multi-Agent Motion Planning With Non-Parametric Closed-Loop Behavior Learning
• An ANN-Enhanced Approach for Flatness-Based Constrained Control of Nonlinear Systems
• Online-Learning Deep Neuro-Adaptive Dynamic Inversion Controller for Model Free Control
• YANNs: Y-wise Affine Neural Networks for Exact and Efficient Representations of Piecewise Linear Functions
• On ANN-enhanced positive invariance for nonlinear flat systems