ARDrone navigation using ePFC controller

Project Overview

This project involved reviewing and implementing an Extended Potential Field Controller (ePFC) for aerial robots, specifically quadcopters. The goal was to assess the controller’s efficiency in guiding drones through dynamic environments while avoiding obstacles, providing an improved method over traditional potential field controllers (PFC).

Importance and Applications

Autonomous aerial robots have seen increased adoption across sectors including emergency response, delivery services, and military operations. These tasks often involve complex environments where precise tracking, obstacle avoidance, and real-time responsiveness are critical. Due to their agility, drones have become essential in scenarios ranging from search-and-rescue missions to delivery and reconnaissance operations.

Extended Potential Field Controller (ePFC)

The ePFC enhances traditional potential field methods by integrating both the relative positions and velocities of the drone with respect to obstacles and targets. While traditional controllers use positional data alone, leading to delayed reactions, ePFC offers more responsive, smooth, and efficient navigation by adding velocity-dependent terms into the control algorithm.

Simulation and Experimentation

We validated the ePFC controller using MATLAB Simulink for theoretical simulations, and Gazebo combined with Robot Operating System (ROS) for practical experiments. Multiple scenarios involving stationary and moving targets, along with various obstacle configurations, were tested rigorously.

Results and Performance

Simulations and experimental tests clearly demonstrated the superiority of the ePFC over traditional PFC. The drone showed significantly reduced overshoot, shorter paths, improved obstacle avoidance, and quicker stabilization times in reaching designated waypoints.

Lessons Learned

Implementing and comparing these controllers provided deep insights into drone dynamics, stability criteria using Lyapunov theory, and the critical differences between simulated and real-world behaviors (the sim-to-real gap). These insights underline the importance of accurate real-time control algorithms for stable aerial robotics operations.

Future Enhancements

Future developments include optimizing path planning algorithms, integrating more complex dynamic obstacle scenarios, and enhancing real-time perception capabilities for even more robust and autonomous drone navigation.

Conclusion

The implementation and review of the ePFC controller effectively demonstrated its capability as a computationally efficient and highly responsive control strategy, suitable for the growing complexities of real-world drone navigation tasks.