Optimal Parameters
After extensive calibration simulation, the following parameters yielded the best performance:
VFF Guidance Parameters
| Parameter | Optimal Value | Description |
|---|---|---|
| \(Q_{goal}\) | -278.21 | Goal charge strength |
| \(F_{goal, const}\) | 253.05 | Constant attraction to goal |
| \(F_{\text{sat}}\) | 427.02 | Obstacle saturation threshold |
| \(F_{\text{fac,squash}}\) | 235.98 | Post-squashing force factor |
| \(Q_{\text{vertical,obs}}\) | 6.56 | Vertical obstacle charge modifier |
Velocity PID Parameters
| Parameter | Optimal Value | Description |
|---|---|---|
| \(K_x, K_y, K_z\) | 0.073, 0.391, 0.583 | Velocity P gains (x, y, z) |
| \(K_{i,x}, K_{i,y}, K_{i,z}\) | 0.00022, 0.091, 0.066 | Velocity I gains (x, y, z) |
| \(K_{d,x}, K_{d,y}, K_{d,z}\) | 0.043, 0.024, 0.0089 | Velocity D gains (x, y, z) |
Attitude Controller Parameters
| Parameter | Optimal Value | Description |
|---|---|---|
| MC_ROLL_P | 5.41 | Roll attitude gain |
| MC_PITCH_P | 8.00 | Pitch attitude gain |
| MC_YAW_P | 3.87 | Yaw attitude gain |
Rate Controller Parameters
| Parameter | Optimal Value | Description |
|---|---|---|
| MC_ROLLRATE_P | 1.00 | Roll rate P gain |
| MC_ROLLRATE_I | 0.081 | Roll rate I gain |
| MC_ROLLRATE_D | 0.0013 | Roll rate D gain |
| MC_ROLLRATE_FF | 1.48 | Roll rate feedforward |
| MC_ROLLRATE_K | 0.504 | Roll rate global gain |
| MC_PITCHRATE_P | 1.00 | Pitch rate P gain |
| MC_PITCHRATE_I | 0.055 | Pitch rate I gain |
| MC_PITCHRATE_D | 0.0034 | Pitch rate D gain |
| MC_PITCHRATE_FF | 1.30 | Pitch rate feedforward |
| MC_PITCHRATE_K | 0.50 | Pitch rate global gain |
| MC_YAWRATE_P | 1.00 | Yaw rate P gain |
| MC_YAWRATE_I | 0.070 | Yaw rate I gain |
| MC_YAWRATE_D | 0.0024 | Yaw rate D gain |
| MC_YAWRATE_FF | 1.21 | Yaw rate feedforward |
| MC_YAWRATE_K | 0.516 | Yaw rate global gain |
Demonstration Videos
C++ Simulation
Drone navigating in a simulated C++ world.
ROS/Gazebo Simulation
PX4 navigation in Gazebo world
Future Work
Based on our results, we plan to focus on:
- Integration with real-world drone hardware at CITI-USP
- Extension to multi-drone coordination scenarios
- Adaptive parameter tuning for different environment types
- Integration with higher-level path planning algorithms
- Performance optimization for embedded systems