"Reinforcement Learning-Based Microrobot Navigation Along Circular Paths Using Magnetic Field Control"

---

🚀 Features

• ✅ Microrobot environment with magnetic field actuation
• ✅ State, Action, Reward, Next State, Done (MDP)
• ✅ Continuous control with Soft Actor-Critic (SAC)
• ✅ Reward shaping to enforce circular trajectory
• ✅ Visual plots of path and performance

---

🤖 Microrobot Navigation using Reinforcement Learning (Soft Actor-Critic)

This project explores autonomous microrobot navigation from one point to another along a circular path, under the influence of a controlled magnetic field, using Deep Reinforcement Learning. It applies the Soft Actor-Critic (SAC) algorithm in a continuous control setting where:

• 📸 The state space is image-based, representing the simulation graph (e.g., robot position on the field).
• 🌀 The action space controls rotation around and along the axis via angle Φ (phi)—simulating torque or field orientation.

---

📷 Visuals

🧭 Circular Navigation Path

📈 Reward Curve over Training

---

🧠 Algorithm: Soft Actor-Critic (SAC)

• Entropy-regularized RL algorithm for stable training
• Continuous action space (perfect for magnetic field control)
• Automatic temperature tuning