Abstract:
Mangrove ecosystems face increasing threats from climate change, anthropogenic activity, and
inadequate monitoring. Effective conservation and management require frequent, large-scale data
collection across complex terrain characterized by dense root structures, tidal inundation, soft
substrates, and GPS-denied environments under dense canopy cover. This study presents the design and
implementation of a sensor integration framework for autonomous rover navigation in mangrove
ecosystem monitoring. The navigation system integrates a BNO055 nine-axis Inertial Measurement
Unit (IMU) with quadrature wheel encoders on an STM32 Nucleo U545RE-Q microcontroller, with
firmware developed and deployed using the MATLAB/Simulink Embedded Coder Support Package
for STM32. Sensor fusion is achieved through a dead-reckoning approach, in which real-time
orientation estimates from the IMU (heading, roll, and pitch) are combined with rotational velocity
measurements (RPM) from the wheel encoders to estimate the rover's relative position and heading in
GPS-denied environments. Simulink models for both sensors were successfully developed, deployed,
and validated, demonstrating reliable data acquisition within expected operating ranges: the encoder
model produced stable RPM readings, and the IMU outputs remained within manufacturer-specified
Euler angle ranges. These results demonstrate the feasibility of a low-power, embedded sensor fusion
platform for autonomous navigation in challenging field environments such as mangrove ecosystems.
The proposed framework provides a foundational sensor layer for future integration with LiDAR-based
Simultaneous Localization and Mapping (SLAM) algorithms and a Robot Operating System (ROS),
enabling fully autonomous environmental monitoring. This work contributes to a replicable embedded
systems framework for developing cost-effective, GPS-independent, terrain-adaptive robotic platforms
for next-generation mangrove ecosystem monitoring.
