2D top-down maps are commonly used for the navigation and exploration of mobile robots through unknown areas. Typically, the robot builds the navigation maps incrementally from local observations using onboard sensors. Recent works have shown that predicting the structural patterns in the environment through learning-based approaches can greatly enhance task efficiency. While many such works build task-specific networks using limited datasets, we show that the existing foundational vision networks can accomplish the same without any fine-tuning. Specifically, we use Masked Autoencoders, pre-trained on street images, to present novel applications for field-of-view expansion, single-agent topological exploration, and multi-agent exploration for indoor mapping, across different input modalities. Our work motivates the use of foundational vision models for generalized structure prediction-driven applications, especially in the dearth of training data. We share more qualitative results at https://raaslab.org/projects/MIM4Robots.

Diabetes is one of the significant causes of blindness, especially among aged adults. As diabetes increases, the vision starts to deteriorate, which is called as diabetic retinopathy. The longer a person has diabetes and the less the blood sugar is controlled, the higher chance the person has diabetic retinopathy. 7.7 million people of age 40 or more have diabetic retinopathy. More than 90% of the cases causing vision loss can be avoided if detected early. Here, we use a deep-learning approach to automatically classify the fundus images into normal, mild, moderate, severe, and proliferative diabetic retinopathy. The dataset is collected from Aravind Eye Care System, and comprises of more than three thousand images. The images in the dataset are captured under diverse illumination conditions. We use a densely connected convolutional neural network architecture for the classification and detection of the severity level. Various pre-processing strategies are applied using OpenCV and Keras library to remove the noise from the image dataset. The model performs training for several epochs, and the hyperparameters are tuned to maximize the performance. Experimental results show that the model can successfully detect the severity of diabetic retinopathy.