IBVS

Image-Based Visual Servoing (IBVS) uses image-space features directly in the feedback loop.

Table of contents

1. Core idea
2. Typical advantages
3. Typical challenges

---

Core idea

1. Detect visual features in the image plane.
2. Define image error relative to desired feature positions.
3. Use an image Jacobian (interaction matrix) to compute camera/robot velocity commands.

Typical advantages

• Directly uses image measurements
• Less dependence on full 3D pose estimation
• Effective for feature-tracking tasks

Typical challenges

• Interaction matrix can become singular
• Sensitive to feature loss and noise
• May yield unintuitive Cartesian trajectories