Undergraduate Thesis – *"Extending AnyPlace Toward Unified Grasp and Placement for Generalized Robotic Manipulation"*

I designed an end-to-end robotic pick-and-place system that unifies grasping and placement into a single diffusion-based visuomotor policy. I redesigned a traditional “grasp-then-place” pipeline into a large-scale demonstration collection system in Isaac Lab, using motion planning and simulation to automatically generate datasets of successful trajectories. These demonstrations were then used to train a RoboMimic diffusion policy that maps raw scene observations directly to feasible, task-aware action sequences, removing the need for hand-tuned pre-placement poses and brittle heuristic planning.