Biorobotic Systems and Biosensors Operated via Machine Learning Techniques, With Application to Underwater Robots and Environmental Monitoring

Bioinspired and biomimetic robots can move efficiently, with agility, even in unstructured scenarios, being highly adaptive and reducing ecosystem disturbance during exploration and inspection tasks. These robots can be applied in agriculture and animal farming, as well as wildlife monitoring. This thesis presents the design and development of aquatic biomimetic robots with a focus on three main areas: 1) the reproducibility of marine-inspired locomotion mechanisms, 2) bioinspired control strategies for enhanced maneuverability and adaptability to diverse aquatic conditions, 3) an integrated sensor system. This sensor system includes proprioceptive sensors, which monitor the robot's internal states, and exteroceptive sensors—such as cameras—that gather external environmental data, enabling both self-awareness and interaction with the surrounding aquatic ecosystem. This work also investigates the use of state-of-the-art machine learning algorithms for analyzing animal behavior, using image and video data to identify distinct behavioral traits based on motor activity patterns. The goal is to leverage these observed individuals as "biosensors" that provide insights into their surrounding environment. These algorithms are developed with the flexibility to be integrated into the bioinspired robotic platforms created within this research, offering a powerful tool for environmental monitoring across different contexts. The synergy between bioinspired robotic systems and behavior analysis algorithms has the potential to advance our understanding of ecosystems while also providing versatile solutions for ecological monitoring and preservation.