Researchers have developed a new model called ManiCM for real-time robotic manipulation. The model uses a consistency constraint to improve efficiency and reduce the number of steps required for denoising in high-dimensional observations. This is achieved by directly predicting robot actions from point cloud inputs, rather than predicting noise. The authors claim that their approach accelerates state-of-the-art methods by 10 times on average while maintaining competitive success rates. --- Why it matters: This matters to researchers and engineers working on robotic manipulation because it has the potential to significantly improve the efficiency of real-time control systems, allowing for faster and more accurate robot actions. This could have implications for applications such as assembly lines, warehouse automation, and search and rescue missions. Source: https://arxiv.org/abs/2406.01586