Krishna Murthy is a PhD candidate at the Robotics and Embodied AI lab and Mila at the University of Montreal, working on differentiable adaptations of physical processes (computer vision, robotics, graphics, physics, and optimization) and their applicability in modern learning pipelines. His work has been recognized with an NVIDIA graduate fellowship (2021) and a best paper award from Robotics and Automation letters (2019). He was also chosen to the RSS Pioneers cohort in 2020.
3D scene graphs (3DSGs) are an emerging description in the robotics and computer vision communities; unifying symbolic, topological, and metric scene representations. However, typical 3DSGs contain hundreds of objects and symbols even for small environments; rendering task planning on the full graph impractical. In this talk, I will present our findings from building the Taskography benchmark, a large-scale robotic task planning benchmark over 3DSGs. We note that neither classical nor learning-based planners are capable of real-time planning over full 3DSGs. Enabling real-time planning thus demands progress on both (a) sparsifying 3DSGs for tractable planning and (b) designing planners that better exploit 3DSG hierarchies. I will present two techniques that draw from (a) and (b) and result in fast task planning over large scene graphs and in long-horizon tasks.