Pathfinder MPC: Charting the Way - Virtual Expo 2025

Welcome to Pathfinder MPC, a project exploring the future of autonomous navigation through a simulated Mars rover. This repository showcases a comparative study between Pure Pursuit Control (PPC) and Model Predictive Control (MPC) using MATLAB simulations. The objective? To identify the best control logic for autonomous rover navigation in dynamic, obstacle-laden 2D environments.

🔗 GitHub Repository

---

Project Abstract

Autonomous navigation is crucial for the next generation of planetary rovers. This project simulates two control strategies—Pure Pursuit Control (PPC) and Model Predictive Control (MPC)—to guide a rover along a path while avoiding static and dynamic obstacles.

• PPC: Lightweight and simple but struggles in complex terrains.
• MPC: Robust, adaptive, and precise, albeit computationally intensive.

By evaluating these methods on path tracking accuracy, computational cost, and obstacle avoidance, this project lays the groundwork for smarter space exploration.

---

Aim

To compare PPC and MPC for rover navigation in a 2D simulated environment by evaluating:

• Path tracking accuracy
• Obstacle avoidance
• Computational efficiency
• Scalability for complex environments

---

Technologies & Models Used

• MATLAB – Simulation environment
• Pure Pursuit Control (PPC) – Geometric path tracking algorithm
• Model Predictive Control (MPC) – Optimization-based predictive controller
• Bicycle Model (BM) – Kinematic model for vehicle motion

---

Methodology

1. PPC Implementation

• Path defined as a smooth sinusoidal curve
• Static circular obstacles included
• Uses a lookahead point to calculate steering angle
• Lightweight but not robust to dynamic or tight obstacle navigation

2. MPC Implementation

• State vector: [x, y, θ, v]
• Uses a prediction horizon to forecast motion
• Solves an optimization problem at each time step with MATLAB’s fmincon
• Penalizes deviation from path, speed spikes, steering changes, and proximity to obstacles
• Adaptable to dynamic obstacles and high-precision tracking

3. Bicycle Model

• Models the rover with a simplified 2-wheel representation
• Captures essential dynamics (position, orientation, velocity, steering angle)

---

Results

Both controllers were tested in identical environments. The simulations visually demonstrate:

• PPC: Simpler, faster, but less effective at obstacle avoidance
• MPC: Smoother paths, higher accuracy, handles dynamic scenarios better

MPC Demo with Dynamic Obstacles

View simulation code

PPC Demo with Dynamic Obstacles

View simulation code

---

📚 Literature & References

1. Model Predictive Control - A Review and Future Directions
2. Pure Pursuit Path Tracking - Adaptive Lookahead
3. Control Systems in Autonomous Robotics
4. The Bicycle Model for Kinematic Path Tracking

---

📁 Folder Structure

MPC-Rover/
├── MPC Model with Dynamic Obstacles.pdf
├── MPC Model Demo.mp4
├── PPC Model.pdf
├── PPC Model Demo.mp4
├── README.md