Trajectory Optimizer

A ROS-independent trajectory optimization library for UAV perching maneuvers.

Overview

This project provides trajectory optimization capabilities for UAV perching, extracted from the original ROS-based Fast-Perching implementation. The library provides:

• TrajOpt: The original class interface, cleaned of ROS dependencies for standalone use

Features

• ROS-Independent: Complete removal of ROS dependencies for use in any C++ project
• MINCO Trajectory Generation: Minimum control effort trajectory optimization
• L-BFGS Optimization: Efficient gradient-based optimization
• Collision Avoidance: Built-in collision detection and avoidance constraints
• Dynamic Constraints: Velocity, acceleration, thrust, and angular velocity limits
• Perching Optimization: Specialized for UAV perching maneuvers

Dependencies

• Eigen3: Linear algebra library
• C++14: Minimum C++ standard required

Building

Windows:

mkdir build
cd build
cmake -G "MinGW Makefiles" ..
mingw32-make -j8

Usage

Basic Example

#include "traj_opt.h"

// Create optimizer
traj_opt::TrajOpt optimizer;

// Configure parameters
optimizer.setDynamicLimits(10.0, 10.0, 20.0, 2.0, 3.0, 2.0);
optimizer.setRobotParameters(1.0, 0.3, 0.1, 0.5);
optimizer.setOptimizationWeights(1.0, -1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0);

// Set initial state
Eigen::MatrixXd initial_state(3, 4);
initial_state.col(0) = Eigen::Vector3d(0.0, 0.0, 2.0);  // position
initial_state.col(1) = Eigen::Vector3d(5.0, 0.0, 0.0);  // velocity

// Set target
Eigen::Vector3d target_position(10.0, 0.0, 1.0);
Eigen::Vector3d target_velocity(2.0, 0.0, 0.0);
Eigen::Quaterniond landing_quaternion(1.0, 0.0, 0.0, 0.0);

// Generate trajectory
Trajectory result_trajectory;
bool success = optimizer.generate_traj(
    initial_state, target_position, target_velocity, 
    landing_quaternion, 10, result_trajectory
);

Configuration Methods

Dynamic Limits

optimizer.setDynamicLimits(
    max_velocity,        // m/s
    max_acceleration,    // m/s²
    max_thrust,         // N
    min_thrust,         // N
    max_angular_velocity,    // rad/s
    max_yaw_angular_velocity // rad/s
);

Robot Parameters

optimizer.setRobotParameters(
    velocity_plus,   // Landing velocity offset
    robot_length,    // Robot length (m)
    robot_radius,    // Robot radius (m)
    platform_radius  // Landing platform radius (m)
);

Optimization Weights

optimizer.setOptimizationWeights(
    time_weight,              // Time penalty weight
    velocity_tail_weight,     // Terminal velocity weight
    position_weight,          // Position constraint weight
    velocity_weight,          // Velocity constraint weight
    acceleration_weight,      // Acceleration constraint weight
    thrust_weight,           // Thrust constraint weight
    angular_velocity_weight, // Angular velocity constraint weight
    perching_collision_weight // Collision avoidance weight
);

Code Structure

├── include/
│   ├── traj_opt.h              # Main trajectory optimizer interface
│   ├── minco.hpp               # MINCO trajectory generation
│   ├── lbfgs_raw.hpp          # L-BFGS optimization
│   ├── poly_traj_utils.hpp    # Polynomial trajectory utilities
│   └── root_finder.hpp        # Root finding utilities
├── src/
│   └── traj_opt_perching.cc   # Implementation
├── examples/
│   ├── basic_example.cpp      # Basic usage example
│   └── precision_test.cpp     # Precision testing
├── scripts/
│   ├── visualize_trajectories.py  # Trajectory visualization
│   └── trajectory_summary.py      # Trajectory analysis
└── CMakeLists.txt             # Build configuration

Key Features

• ROS-Independent: Complete elimination of ros::NodeHandle and ROS-specific code
• Clean Interface: Simplified API with clear setter methods
• Visualization Tools: Python scripts for trajectory analysis and visualization
• Example Code: Working examples demonstrating usage
• MINCO Optimization: Minimum control effort trajectory generation

Original Attribution

This code is derived from the Fast-Perching project:

• Repository: ZJU-FAST-Lab/Fast-Perching
• Original Authors: ZJU FAST Lab

License

Please refer to the original Fast-Perching repository for licensing information.