Control Systems
Machine Learning
Deep Learning
Autonomous Underwater Vehicle
Navigation in Autonomous Systems
Modelling and Control of AUV:
AUV modelling and control focuses on designing robust systems that enable Autonomous Underwater Vehicles (AUVs) to navigate complex and unpredictable marine environments. These environments are characterized by factors such as fluctuating water currents, depth variations, temperature gradients, and salinity changes, all of which can significantly affect the stability and performance of AUVs. Additionally, AUVs often encounter uncertainties and unmodeled dynamics that make precise control a challenge. To address these issues, I employ advanced control techniques like Model Predictive Control (MPC) and Control Barrier Functions (CBF). MPC allows for dynamic adjustments in control strategies, ensuring stability while optimizing energy consumption. CBFs provide additional robustness, ensuring safe operation by preventing undesired behavior in the face of external disturbances or internal system uncertainties. These approaches are critical for ensuring the reliability and efficiency of AUVs, especially during prolonged underwater missions where energy conservation is crucial.
AUV Navigation:
In the realm of AUV navigation, focus is on developing precise and resilient systems that allow AUVs to traverse vast and often uncharted underwater terrains. This includes environments where traditional navigation aids such as Doppler Velocity Loggers (DVLs) are unavailable or unreliable, known as DVL-denied conditions. Navigational precision in these environments is vital for several applications, including ocean floor mapping, environmental monitoring, disaster response, and the autonomous exploration of remote underwater ecosystems. Innovative solutions leverage alternative sensors, such as inertial measurement units (IMUs), pressure sensors, and acoustic positioning systems, to create accurate localization and path planning capabilities. These systems allow AUVs to operate effectively in deep-sea environments, providing valuable data for scientific research, resource exploration, and national security.
Underwater Perception:
a. Image Enhancement: Underwater imaging poses several challenges due to the unique properties of the aquatic environment, including light scattering, absorption, and turbidity, which significantly degrade image quality. Traditional imaging techniques are often ineffective in capturing clear, accurate visuals of underwater objects, which can hinder mission success. To overcome these limitations, deep learning models are being utilized for advanced image enhancement in AUV systems. These models are designed to restore clarity, adjust color balance, and compensate for distortion caused by the underwater medium, resulting in clearer and more accurate visual representations of the environment. Applications of these image enhancement techniques are particularly valuable in tasks such as coral reef monitoring, where maintaining an accurate representation of reef health is critical, or in artifact detection for archaeological surveys. By improving image quality, AUVs can identify and document marine ecosystems, historical shipwrecks, and other underwater features with greater detail and reliability.
b. Object Detection:
In addition to enhancing image quality, AUVs are increasingly equipped with object detection capabilities that allow them to identify and classify underwater objects in real-time. Using advanced sonar-based imaging systems combined with machine learning algorithms, AUVs can detect a wide range of objects, including marine life, shipwrecks, underwater mines, and other submerged hazards. This capability is essential in low-visibility environments, where traditional optical imaging may be insufficient. Object detection algorithms are designed to process sonar data and interpret complex underwater scenes, enabling the AUV to identify objects, assess their size, shape, and location, and classify them according to their relevance to the mission. This enhances the AUV’s ability to perform complex tasks such as underwater inspections, search-and-rescue operations, and environmental monitoring. For example, during an underwater inspection