Pointgoal Navigation using Hierarchical Reinforcement Learning

Navigation and localisation are especially important from the perspective of robotic vision. In this project, we leverage learning-based approaches to improve these tasks and address the limitations of classical approaches. Image synthesis with Generative Adversarial Networks (GAN) is an active area of research, where improving GAN instability can prove crucial for producing strong estimates for distributions. In this project, we target and subsequently utilise these distributions for improving other tasks related to computer vision.