Mobile Robotic Platform for Multi-Disease Retinal Screening : A Comprehensive Review

Retinal diseases such as diabetic retinopathy, age-related macular degeneration, glaucoma, cataract-related retinal changes and retinal vascular disorders remain major causes of preventable vision loss. However, this problem is more serious in rural and resource-constrained regions because screening still depends on trained specialists, manual fundus interpretation and hospital-based imaging systems. Recent studies show that deep learning, vision transformers, hybrid CNN-transformer models and multimodal systems can classify retinal diseases with high accuracy. Yet, many models are designed for offline testing or single-device diagnosis. They do not fully solve the field-level screening problem where tasks like image capture, patient alignment, local inference and remote consultation must work together. Therefore, this review suggests the need for an autonomous robotic retinal screening system which combines non-mydriatic fundus imaging, edge-AI inference and tele-ophthalmology support. The proposed framework uses CNN modules for local lesion features, transformer modules for global retinal context and optional multimodal fusion with physiological inputs. Model compression is done through quantisation, and pruning is considered for real-time edge deployment. Thus, the review connects recent retinal AI advances with the practical engineering pathway for scalable multi-disease screening in clinics, outreach camps and underserved healthcare settings.