Contact Estimation

Contact estimation, the process of determining where and how objects are in contact, is crucial for robotics and human-computer interaction, aiming to enable more robust and dexterous manipulation and interaction. Current research focuses on developing accurate and efficient methods using diverse data sources (visual, auditory, tactile) and model architectures, including diffusion models, neural networks (e.g., VAEs, graph-based networks), and particle filters, often combined with classical methods like trajectory optimization. These advancements are significantly impacting fields like robotic manipulation, augmented/virtual reality, and human motion capture by enabling more realistic simulations and improved control of interactions with the environment.