OpEn (Optimization Engine) is a C++ library and code generator for fast, embedded, nonconvex optimization, targeting robotics and autonomous systems. It allows users to define parametric optimization problems and generate highly efficient, self-contained Rust code for their solvers, enabling real-time performance on embedded hardware.

How It Works

OpEn leverages a custom numerical algorithm implemented in Rust for solving nonconvex parametric optimization problems. It generates parametric optimizer modules in Rust, which can be compiled into standalone libraries or run as a TCP service. This approach prioritizes speed, memory safety, and deployability on resource-constrained embedded devices, abstracting away the complexities of numerical optimization for application developers.

Quick Start & Requirements

• Installation: Primarily through code generation via Python or MATLAB interfaces.
• Prerequisites: Python 3.x, MATLAB, Rust, and CasADi. ROS support is available.
• Resources: Embedded applications can run on devices like Intel Atom with low CPU usage (e.g., 15% CPU at 20Hz).
• Documentation: alphaville.github.io/optimization-engine

Highlighted Details

• Generates Rust code for parametric optimizers, enabling fast, safe, and embedded execution.
• Supports nonconvex parametric optimization problems of a general form.
• Offers auto-generation of ROS packages for seamless integration into robotic systems.
• Demonstrated real-time performance on embedded hardware for autonomous navigation.

Maintenance & Community

• Core team includes Pantelis Sopasakis and Emil Fresk.
• Community discussion channels available on Discord and Gitter.
• Project encourages contributions via established guidelines.

Licensing & Compatibility

• Dual-licensed under Apache License v2.0 or MIT License.
• Permissive licenses suitable for commercial use and integration into closed-source projects.

Limitations & Caveats

The README focuses on code generation and embedded performance; details on the specific range of supported nonconvex problem structures or potential numerical stability issues for highly challenging problems are not extensively detailed.