Saravanabalagi is a Computer Vision PhD student working under the supervision of Dr. John McDonald at Maynooth University on long term autonomy for self driving vehicles. The objective of his research is to build novel approaches to place recognition and image based localisation in challenging conditions. The project is funded through Lero Irish Software Research Centre and is being carried out in collaboration with Valeo Vision Systems.

Saravanabalagi focuses on formulating a novel approach to place recognition using a combination of CNNs and LSTMs to generate compact embeddings that encode semantic, visual and temporal features, thereby providing greater robustness to variations caused due to different times of the day, weather, lighting and seasons.

Github: https://github.com/saravanabalagi

Publications

2023

Horgan, Jonathan, McDonald, John, Ramachandran, Saravanabalagi, Sistu, Ganesh. 2023. Scalable and Efficient Hierarchical Visual Topological Mapping. In: 2023 21st International Conference on Advanced Robotics (ICAR), IEEE. doi: 10.1109/icar58858.2023.10406394.

2022

Horgan, J, McDonald, J, Ramachandran, S, Sistu, G. 2022. Fast and Efficient Scene Categorization for Autonomous Driving using VAEs. In: Irish Machine Vision and Image Processing Conference.

2019

McDonald, J, Ramachandran, S. 2019. Place Recognition in Challenging Conditions. In: Irish Machine Vision and Image Processing Conference. doi: 10.21427/seen-pq19.