Search

Contrastive learning has recently narrowed the gap between self-supervised and supervised methods in image and video domain. State-of-the-art video contrastive learning methods such as CVRL and $\rho$-MoCo spatiotemporally augment two clips from the same video as positives. By only sampling positive clips locally from a single video, these methods neglect other semantically related videos that can also be useful. To address this limitation, we leverage nearest-neighbor videos from the global space as additional positive pairs, thus improving positive key diversity and introducing a more relaxed notion of similarity that extends beyond video and even class boundaries. Our method, Inter-Intra Video Contrastive Learning (IIVCL), improves performance on a range of video tasks., Comment: Accepted to the ICLR 2023 Workshop on Mathematical and Empirical Understanding of Foundation Models

Multi-goal reinforcement learning (RL) aims to qualify the agent to accomplish multi-goal tasks, which is of great importance in learning scalable robotic manipulation skills. However, reward engineering always requires strenuous efforts in multi-goal RL. Moreover, it will introduce inevitable bias causing the suboptimality of the final policy. The sparse reward provides a simple yet efficient way to overcome such limits. Nevertheless, it harms the exploration efficiency and even hinders the policy from convergence. In this paper, we propose a density-based curriculum learning method for efficient exploration with sparse rewards and better generalization to desired goal distribution. Intuitively, our method encourages the robot to gradually broaden the frontier of its ability along the directions to cover the entire desired goal space as much and quickly as possible. To further improve data efficiency and generality, we augment the goals and transitions within the allowed region during training. Finally, We evaluate our method on diversified variants of benchmark manipulation tasks that are challenging for existing methods. Empirical results show that our method outperforms the state-of-the-art baselines in terms of both data efficiency and success rate., Comment: 8 pages, 7 figures

Despite the impressive progress achieved in robotic grasping, robots are not skilled in sophisticated tasks (e.g. search and grasp a specified target in clutter). Such tasks involve not only grasping but the comprehensive perception of the world (e.g. the object relationships). Recently, encouraging results demonstrate that it is possible to understand high-level concepts by learning. However, such algorithms are usually data-intensive, and the lack of data severely limits their performance. In this paper, we present a new dataset named REGRAD for the learning of relationships among objects and grasps. We collect the annotations of object poses, segmentations, grasps, and relationships for the target-driven relational grasping tasks. Our dataset is collected in both forms of 2D images and 3D point clouds. Moreover, since all the data are generated automatically, it is free to import new objects for data generation. We also released a real-world validation dataset to evaluate the sim-to-real performance of models trained on REGRAD. Finally, we conducted a series of experiments, showing that the models trained on REGRAD could generalize well to the realistic scenarios, in terms of both relationship and grasp detection. Our dataset and code could be found at: https://github.com/poisonwine/REGRAD

Nitrogen doping study on niobium (Nb) samples used for the fabrication of superconducting radio frequency (SRF) cavities was carried out. The samples' surface treatment was attempted to replicate that of the Nb SRF cavities, which includes heavy electropolishing (EP), nitrogen doping and the subsequent EP with different amounts of material removal. The surface chemical composition of Nb samples with different post treatments has been studied by XPS. The chemical composition of Nb, O, C and N was presented before and after Gas Cluster Ion Beam (GCIB) etching. No signals of poorly superconducting nitrides NbNx was found on the surface of any doped Nb sample with the 2/6 recipe before GCIB etching. However, in the depth range greater than 30nm, the content of N element is below the XPS detection precision scope even for the Nb sample directly after nitrogen doping treatment with the 2/6 recipe., Comment: 14 pages, 18 figures

Nitrogen doping treatment with the subsequent electropolishing (EP) of the niobium superconducting cavity can significantly increase the cavity's quality factor up to a factor of 3. But the process of the EP removal may reintroduce hydrogen in the cavity surface, which may influence the cavity's radio frequency performance. Plasma cleaning study on niobium samples with gas mixtures of argon and oxgen intended to remove contaminations (hydrocarbons and micronicdust particles) from cavity surface to avoid field emission, was performed in Peking University. The niobium samples have been analyzed using the time of flight secondary ion mass spectrometry (TOF-SIMS) to measure the depth profiles of H, C, O, F, P and Nb. The measuring results show that the plasma cleaning with gas mixtures of argon and oxgen and conditions of about 20Pa and 100W can remarkably reduce the contents of impurity elements in the depth of about 30 nm without introducing hydrogen in the cavity surface. So plasma cleaning has been proposed to be a new possible treatment for niobium superconducting cavity after nitrogen doping.