Conic IPDDP

Overview

C++ implementation of the Conic Interior Point Differential Dynamic Programming (Conic IPDDP), designed for solving connic-constrained optimization problems.

$$ \min_{u_{0:T-1}} \ell_f(x_T) + \sum_{t=0}^{T-1} \ell_t(x_t, u_t) $$

$$ \mathrm{subject \ to } \quad x_{t+1} = f_t(x_t, u_t), \quad x_0 = x_{\text{init}} $$

$$ -g_t(x_t, u_t) \in K_1, \quad -h_t(x_t, u_t) \in K_2 $$

$$ -g_T(x_T) \in K_1, \quad -h_T(x_T) \in K_2 $$

$$ \quad t = 0,1,\dots,T-1 $$

Feature

• Automatic conversion between different optimization frameworks:

  • unconstrained DDP
  • Interior Point DDP
  • Conic IPDDP

• Supports multiple constraints

• Provides automatic differentiation

• Allows user-defined function overrides

Installation

Follow these steps to install and set up the project:

# Clone the repository
git clone https://github.com/MC00614/ipddp_cpp.git

# Navigate to the project directory
cd ipddp_cpp

# Install dependencies

# 1. Eigen 3.3.9
wget https://gitlab.com/libeigen/eigen/-/archive/3.3.9/eigen-3.3.9.tar.gz
tar -xvzf eigen-3.3.9.tar.gz
mv eigen-3.3.9 eigen

# 2. autodiff 1.0.0
wget https://github.com/autodiff/autodiff/archive/refs/tags/v1.0.0.tar.gz
tar -xvzf v1.0.0.tar.gz 
mv autodiff-1.0.0 autodiff

# 3. matplotlibcpp
git clone https://github.com/lava/matplotlib-cpp.git

Usage

• Include the solver header and compile it together with your project.
• Define your model by inheriting ModelBase.

Build & Run

# Create and navigate to the build directory
mkdir build && cd build

# Run CMake and compile the project
cmake ..
make -j$(nproc)

# Execute the solver
./ipddp

Examples

To build with example cases:

cmake .. -DBUILD_EXAMPLE=ON
make -j$(nproc)

2D Rocket

# Run 2D Rocket example
./rocket2d

3D Drone

# Run 3D drone example
./drone3d