CHOMP

Table of contents

1. Core idea
2. Why CHOMP for manipulation
3. Practical notes

---

CHOMP (Covariant Hamiltonian Optimization for Motion Planning) is a trajectory optimization method that improves an initial path into a smooth, collision-free motion.

Core idea

1. Start from an initial trajectory guess.
2. Define a cost that combines smoothness and obstacle avoidance.
3. Iteratively optimize the trajectory by gradient-based updates.

Why CHOMP for manipulation

• Produces smooth, executable trajectories for arm motion.
• Handles cluttered environments through obstacle-cost gradients.
• Integrates well with kinematic and collision constraints.

Practical notes

• Quality depends on the initial trajectory.
• Good signed-distance fields improve obstacle gradients.
• Often used with post-processing or time-parameterization.