Abstract:Conic constraints appear in many important control applications like legged locomotion, robotic manipulation, and autonomous rocket landing. However, current solvers for conic optimization problems have relatively heavy computational demands in terms of both floating-point operations and memory footprint, making them impractical for use on small embedded devices. We extend TinyMPC, an open-source, high-speed solver targeting low-power embedded control applications, to handle second-order cone constraints. We also present code-generation software to enable deployment of TinyMPC on a variety of microcontrollers. We benchmark our generated code against state-of-the-art embedded QP and SOCP solvers, demonstrating a two-order-of-magnitude speed increase over ECOS while consuming less memory. Finally, we demonstrate TinyMPC's efficacy on the Crazyflie, a lightweight, resource-constrained quadrotor with fast dynamics. TinyMPC and its code-generation tools are publicly available at

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve The paper aims to address the computational and memory bottlenecks encountered when executing real-time Model Predictive Control (MPC) on resource-constrained microcontrollers. Specifically, the paper focuses on how to efficiently handle optimization problems with Second-Order Cone (SOC) constraints on these microcontrollers. ### Main Problem Background 1. **Importance of Second-Order Cone Constraints**: - Second-order cone constraints appear in many important control applications, such as legged robot locomotion, robotic manipulation, and autonomous rocket landing. These constraints are often used to describe physical characteristics like friction, posture, and thrust limits. 2. **Limitations of Existing Solvers**: - Current second-order cone optimization solvers have relatively high computational demands, including floating-point operations and memory usage, making them impractical for small embedded devices. - Existing solvers typically do not support warm starts, failing to fully utilize the computational results from previous time steps, leading to poor performance. - Most solvers are not optimized for embedded systems and cannot run efficiently on resource-constrained microcontrollers. 3. **Need for TinyMPC**: - TinyMPC is an open-source high-performance solver designed for low-power embedded control applications. The paper extends TinyMPC to handle second-order cone constraints and provides code generation tools for deployment on various microcontrollers. ### Solution 1. **Extension of TinyMPC**: - The paper extends TinyMPC to handle second-order cone constraints on states and inputs, enabling real-time second-order cone model predictive control. - By precomputing and caching expensive matrix decompositions, TinyMPC reduces online computational load and improves computational speed. 2. **Code Generation Tools**: - The paper develops an open-source code generation package with interfaces for Python, MATLAB, and Julia, making it easy for users to deploy TinyMPC on various microcontrollers. 3. **Performance Validation**: - The paper demonstrates TinyMPC's performance advantages on ARM Cortex M4 and M7 microcontrollers through benchmark tests and hardware experiments, particularly in terms of memory usage and computational speed. - Experiments include multiple control tasks on the Crazyflie nano quadrotor drone, validating the effectiveness and practicality of TinyMPC. ### Main Contributions 1. **Open-Source Implementation**: - Provides an open-source implementation of TinyMPC supporting second-order cone constraints, making it the fastest low-memory embedded SOCP solver. 2. **User-Friendly Interfaces**: - Offers easy-to-use code generation and solution validation interfaces, supporting Python, MATLAB, and Julia. 3. **Experimental Validation**: - Conducts hardware experiments on multiple benchmark control problems to validate the effectiveness of TinyMPC. Through these improvements, TinyMPC can efficiently execute model predictive control with second-order cone constraints on resource-constrained microcontrollers, providing a new solution for real-time control in embedded systems.

Code Generation for Conic Model-Predictive Control on Microcontrollers with TinyMPC

TinyMPC: Model-Predictive Control on Resource-Constrained Microcontrollers

Embedded MPC Controller Based on Interior‐Point Method with Convergence Depth Control

Tiny Learning-Based MPC for Multirotors: Solver-Aware Learning for Efficient Embedded Predictive Control

Embedded Model Predictive Controller on Low-Cost Low-End Microcontroller for Electrical Drives

Fast Second-order Cone Programming for Safe Mission Planning

Neuromorphic quadratic programming for efficient and scalable model predictive control

Embedded Model Predictive Control Using Robust Penalty Method

Parallel and Proximal Constrained Linear-Quadratic Methods for Real-Time Nonlinear MPC

Embedded Hierarchical MPC for Autonomous Navigation

Fast Contact-Implicit Model Predictive Control

Field Programmable Gate Array/system on a Programmable Chip-Based Implementation of Model Predictive Controller

Chance-Constrained Convex MPC for Robust Quadruped Locomotion Under Parametric and Additive Uncertainties

Efficiently solving the harmonic model predictive control formulation

A sparse ADMM-based solver for linear MPC subject to terminal quadratic constraint

Real-Time Model Predictive Control for Industrial Manipulators with Singularity-Tolerant Hierarchical Task Control

A Parallel-in-Time Newton's Method for Nonlinear Model Predictive Control

PID-Based Event-Triggered MPC for Constrained Nonlinear Cyber-Physical Systems: Theory and Application

Practical Offset-Free Model Predictive Control and Its Embedded Application to Aeroengines

Efficient Calibration of Embedded MPC

Smooth Computation without Input Delay: Robust Tube-Based Model Predictive Control for Robot Manipulator Planning