Multimodal Deep Learning Framework for Detection and Autonomous Removal of Submerged Ocean Plastic Waste

Abstract Marine plastic pollution represents a persistent environmental challenge, particularly when plastic waste becomes submerged, trapped in marine habitats, or deposited on the seabed where conventional monitoring and clean-up methods become difficult. This chapter presents a multimodal deep learning framework for the detection, localisation, classification, and autonomous removal of submerged plastic waste. The framework integrates optical imaging, sonar and acoustic sensing, depth information, deep learning, and autonomous marine robotics. Convolutional neural networks, object-detection models including YOLO and Faster R-CNN, semantic segmentation approaches including U-Net and DeepLab, and reinforcement learning are incorporated into a perception-to-action pipeline for AUV and ROV deployment. The research design addresses underwater visibility, multimodal data fusion, limited annotated data, three-dimensional localisation, lightweight onboard inference, autonomous navigation, grasp planning, and environmentally sensitive retrieval. The framework provides a structured pathway towards scalable AI-supported monitoring and removal of submerged plastic waste while supporting long-term marine environmental sustainability. The proposed framework supports Sustainable Development Goal 14 (Life Below Water), particularly Target 14.1, by advancing AI-driven monitoring and potential removal of submerged plastic waste.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-12
DOI
https://doi.org/10.21203/rs.3.rs-11197394/v1
Primary Topic
Microplastics and Plastic Pollution
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Multimodal Deep Learning Framework for Detection and Autonomous Removal of Submerged Ocean Plastic Waste

Suvendu Chandan Nayak, Shahen Ramnath, Sreedevi Vallabapurapu
Research Square
Microplastics and Plastic Pollution
preprint

Multimodal Deep Learning Framework for Detection and Autonomous Removal of Submerged Ocean Plastic Waste

Suvendu Chandan Nayak, Shahen Ramnath, Sreedevi Vallabapurapu
preprint en

Abstract

Abstract Marine plastic pollution represents a persistent environmental challenge, particularly when plastic waste becomes submerged, trapped in marine habitats, or deposited on the seabed where conventional monitoring and clean-up methods become difficult. This chapter presents a multimodal deep learning framework for the detection, localisation, classification, and autonomous removal of submerged plastic waste. The framework integrates optical imaging, sonar and acoustic sensing, depth information, deep learning, and autonomous marine robotics. Convolutional neural networks, object-detection models including YOLO and Faster R-CNN, semantic segmentation approaches including U-Net and DeepLab, and reinforcement learning are incorporated into a perception-to-action pipeline for AUV and ROV deployment. The research design addresses underwater visibility, multimodal data fusion, limited annotated data, three-dimensional localisation, lightweight onboard inference, autonomous navigation, grasp planning, and environmentally sensitive retrieval. The framework provides a structured pathway towards scalable AI-supported monitoring and removal of submerged plastic waste while supporting long-term marine environmental sustainability. The proposed framework supports Sustainable Development Goal 14 (Life Below Water), particularly Target 14.1, by advancing AI-driven monitoring and potential removal of submerged plastic waste.

Research Square
University of South Africa (ZA)
Responsible consumption and production, Life below water
Microplastics and Plastic Pollution
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon