The field of object manipulation and simulation is rapidly advancing, with a focus on developing innovative solutions for complex tasks such as knot detection, 3D object rigging, and garment folding. Researchers are exploring the use of machine learning techniques, such as deep learning and reinforcement learning, to improve the accuracy and efficiency of these tasks. Additionally, there is a growing interest in integrating temporal context and physics-driven approaches to enhance the performance of simulation models. Noteworthy papers in this area include:
- Anymate: A Dataset and Baselines for Learning 3D Object Rigging, which presents a large-scale dataset and a learning-based auto-rigging framework for 3D objects.
- WisePanda, a physics-driven deep learning framework for rejoining fragmented ancient bamboo slips, which demonstrates the potential of incorporating physical principles into deep learning models.
- FoldNet: Learning Generalizable Closed-Loop Policy for Garment Folding, which proposes a synthetic garment dataset and a keypoint-driven asset and demonstration synthesis approach for robotic garment folding.