Trials are conducted to evaluate information-theoretic mapping strategies
| Planner | Strategy | Description |
|---|---|---|
| Planner 1 | Pure Mutual Information | Selects the next waypoint solely based on expected information gain (MI). |
| Planner 2 | MI + Distance | Balances information gain with spatial diversity, adding a distance-based exploration term. |
Both planners operate over a simulated 2-D thermal field generated by radial_field.py, representing an underlying Gaussian hotspot distribution.
The goal is to sample the field efficiently and reconstruct it
# 1. Start the PX4 simulation
cd PX4-Autopilot
make px4_sitl_r1_rover
# 2. In a new terminal, launch the MI planner (choose version 1 or 2)
cd ~/workspaces/aquatic-mapping
ros2 launch info_gain mi_planner.launch.py planner:=1 # Pure MI
# or
ros2 launch info_gain mi_planner.launch.py planner:=2 # MI + Distance
RMSE: 0.398 °C
MAE : 0.308 °C
Corr: 0.988
RMSE: 1.084 °C
MAE : 0.564 °C
Corr: 0.935