Showing papers by "Li Fei-Fei published in 2023"

MimicPlay: Long-Horizon Imitation Learning by Watching Human Play

Abstract: Imitation Learning from human demonstrations is a promising paradigm to teach robots manipulation skills in the real world, but learning complex long-horizon tasks often requires an unattainable amount of demonstrations. To reduce the high data requirement, we resort to human play data - video sequences of people freely interacting with the environment using their hands. We hypothesize that even with different morphologies, human play data contain rich and salient information about physical interactions that can readily facilitate robot policy learning. Motivated by this, we introduce a hierarchical learning framework named MimicPlay that learns latent plans from human play data to guide low-level visuomotor control trained on a small number of teleoperated demonstrations. With systematic evaluations of 14 long-horizon manipulation tasks in the real world, we show that MimicPlay dramatically outperforms state-of-the-art imitation learning methods in task success rate, generalization ability, and robustness to disturbances. More details and video results could be found at https://mimic-play.github.io

Chemical Composition, Antibacterial Test, and Antioxidant Activity of Essential Oils from Fresh and Dried Stropharia rugosoannulata

Abstract: The essential oils, respectively, from fresh and dried Stropharia rugosoannulata fruiting bodies, an important edible mushroom, have been studied for their chemical composition, antibacterial capacity, and antioxidant activity. The essential oils were obtained by hydrodistillation and analyzed by gas chromatography-mass spectrometry (GC-MS) combined with Kovats retention index. The oils’ antibacterial test was evaluated by the microdilution method against Escherichia coli, Staphylococcus aureus, and Pseudomonas aeruginosa, and antioxidant activity was determined through DPPH radical scavenging activity and ferric reducing power. Twenty-nine components were identified from the fresh mushroom, and the compositions were mainly dominated by hydrocarbons (54.72%), acids (32.99%), esters (5.07%), and terpenic compounds (0.96%). Thirty-five components were identified from the dried sample, and acids (31.22%), terpenic compounds (28.7%), alcohols (12.7%), and ketones (10.48%) were the major compounds. Strong antibacterial capacity and obvious antioxidant activity were observed for both essential oils from the fresh and dried mushrooms.

The ObjectFolder Benchmark: Multisensory Learning with Neural and Real Objects

Abstract: We introduce the ObjectFolder Benchmark, a benchmark suite of 10 tasks for multisensory object-centric learning, centered around object recognition, reconstruction, and manipulation with sight, sound, and touch. We also introduce the ObjectFolder Real dataset, including the multisensory measurements for 100 real-world household objects, building upon a newly designed pipeline for collecting the 3D meshes, videos, impact sounds, and tactile readings of real-world objects. We conduct systematic benchmarking on both the 1,000 multisensory neural objects from ObjectFolder, and the real multisensory data from ObjectFolder Real. Our results demonstrate the importance of multisensory perception and reveal the respective roles of vision, audio, and touch for different object-centric learning tasks. By publicly releasing our dataset and benchmark suite, we hope to catalyze and enable new research in multisensory object-centric learning in computer vision, robotics, and beyond. Project page: https://objectfolder.stanford.edu

Preparation of Zn−Gly and Se−Gly and Their Effects on the Nutritional Quality of Tea (Camellia sinensis)

Abstract: Background: Micronutrient malnutrition affects millions of people due to a lack of Zn and Se. Methods: The process conditions for the manufacture of glycine−chelated sodium selenite (Se−Gly) and zinc sulfate heptahydrate (Zn−Gly) were studied. The effects of ligand concentration, pH, reaction ratio, reaction temperature, and reaction time on fertilizer stability were assessed. The effects of Zn−Gly and Se−Gly on tea plants were determined. Results: Orthogonal experiments showed that the optimal preparation conditions for Zn−Gly (75.80 % Zn chelation rate) were pH 6.0, ligand concentration 4 %, reaction ratio 1:2, reaction time 120 min, reaction temperature 70 ℃. The optimal preparation conditions for Se−Gly (56.75 % Se chelation rate) were pH 6.0, ligand concentration 10%, reaction ratio 2:1, reaction time 40 min, temperature 50 ℃. Each chelate was completely soluble in water and verified by infrared spectroscopy and ultraviolet spectroscopy. Conclusions: Zn−Gly and Se−Gly increased the Zn and Se content in tea plants, and foliar application was more effective than soil application. Combined application of Zn−Gly and Se−Gly was more effective than Zn−Gly or Se−Gly alone. Our findings suggest that Zn−Gly and Se−Gly provide a convenient method of addressing human Zn and Se deficiency.

Identification of an indole biodegradation gene cluster from Providencia rettgeri and its contribution in selectively biosynthesizing Tyrian purple

Abstract: Tyrian purple, mainly composed of 6, 6′-dibromoindigo, is a precious dye extracted from sea snails. In this study, we found Tyrian purple can be selectively produced by a bacterial strain GS-2 when fed with 6-bromotryptophan in the presence of tryptophan. This GS-2 strain was then identified as Providencia rettgeri based on bacterial genome sequencing analysis. An indole degradation gene cluster for indole metabolism was identified from this GS-2 strain. The heterologous expression of the indole degradation gene cluster in Escherichia coli BL21 (DE3) and in vitro enzymatic reaction demonstrated that the indole biodegradation gene cluster may contribute to selectively biosynthesizing Tyrian purple. To further explore the underlying mechanism of the selectivity, we explored the intermediates in this indole biodegradation pathway using liquid chromatography electrospray ionization quadrupole time-of-flight mass spectrometry (LC-ESI-QTOF-MS/MS), which indicated that the indole biodegradation pathway in Providencia rettgeri is the catechol pathway. Interestingly, the monooxygenase GS-C co-expressed with its corresponding reductase GS-D in the cluster has better activity for the biosynthesis of Tyrian purple compared with the previously reported monooxygenase from Methylophaga aminisulfidivorans (MaFMO) or Streptomyces cattleya cytochrome P450 enzyme (CYP102G4). This is the first study to show the existence of an indole biodegradation pathway in Providencia rettgeri, and the indole biodegradation gene cluster can contribute to the selective production of Tyrian purple.

Identification of long noncoding RNAs involved in plumule-vernalization of Chinese cabbage

Abstract: Vernalization is a phenomenon in which plants must undergo a period of continuous low temperatures to change from the vegetative growth stage to the reproductive growth stage. Chinese cabbage is a heading vegetable, and flowering time is an essential developmental trait. Premature vernalization leads to premature bolting, which causes a loss of product value and yield. While research into vernalization has provided a wealth of information, a complete understanding of the molecular mechanism for controlling vernalization requirements has not yet been elucidated. In this study, using high-throughput RNA sequencing, we analyzed the plumule-vernalization response of mRNA and long noncoding RNA in the bolting-resistant Chinese cabbage double haploid (DH) line ‘Ju Hongxin’ (JHX). A total of 3382 lncRNAs were identified, of which 1553 differentially expressed (DE) lncRNAs were characterized as plumule-vernalization responses. The ceRNA network revealed that 280 ceRNA pairs participated in the plumule-vernalization reaction of Chinese cabbage. Through identifying DE lncRNAs in Chinese cabbage and analyzing anti-, cis-, and trans-functional analysis, some candidate lncRNAs related to vernalization promoting flowering of Chinese cabbage and their regulated mRNA genes were found. Moreover, the expression of several critical lncRNAs and their targets was verified using qRT-PCR. Furthermore, we identified the candidate plumule-vernalization-related long noncoding RNAs that regulate BrFLCs in Chinese cabbage, which was interesting and different from previous studies and was a new discovery. Our findings expand the knowledge of lncRNAs in the vernalization of Chinese cabbage, and the identified lncRNAs provide rich resources for future comparative and functional studies.

Siamese Masked Autoencoders

Abstract: Establishing correspondence between images or scenes is a significant challenge in computer vision, especially given occlusions, viewpoint changes, and varying object appearances. In this paper, we present Siamese Masked Autoencoders (SiamMAE), a simple extension of Masked Autoencoders (MAE) for learning visual correspondence from videos. SiamMAE operates on pairs of randomly sampled video frames and asymmetrically masks them. These frames are processed independently by an encoder network, and a decoder composed of a sequence of cross-attention layers is tasked with predicting the missing patches in the future frame. By masking a large fraction ($95\%$) of patches in the future frame while leaving the past frame unchanged, SiamMAE encourages the network to focus on object motion and learn object-centric representations. Despite its conceptual simplicity, features learned via SiamMAE outperform state-of-the-art self-supervised methods on video object segmentation, pose keypoint propagation, and semantic part propagation tasks. SiamMAE achieves competitive results without relying on data augmentation, handcrafted tracking-based pretext tasks, or other techniques to prevent representational collapse.

Design, Synthesis and Biological Evaluation of Novel Pleuromutilin Derivatives Containing 6-Chloro-1-R-1H-pyrazolo[3,4-d]pyrimidine-4-amino Side Chain

Abstract: Two series of pleuromutilin derivatives were designed and synthesized as inhibitors against Staphylococcus aureus (S. aureus). 6-chloro-4-amino-1-R-1H-pyrazolo[3,4-d]pyrimidine or 4-(6-chloro-1-R-1H-pyrazolo[3,4-d]pyrimidine-4-yl)amino-phenylthiol were connected to pleuromutilin. A diverse array of substituents was introduced at the N-1 position of the pyrazole ring. The in vitro antibacterial activities of these semisynthetic derivatives were evaluated against two standard strains, Methicillin-resistant Staphylococcus aureus (MRSA) ATCC 43300, Staphylococcus aureus (S. aureus), ATCC 29213 and two clinical S. aureus strains (144, AD3) using the broth dilution method. Compounds 12c, 19c and 22c (MIC = 0.25 μg/mL) manifested good in vitro antibacterial ability against MRSA which was similar to that of tiamulin (MIC = 0.5 μg/mL). Among them, compound 22c killed MRSA in a time-dependent manner and performed faster bactericidal kinetics than tiamulin in time–kill curves. In addition, compound 22c exhibited longer PAE than tiamulin, and showed no significant inhibition on the cell viability of RAW 264.7, Caco-2 and 16-HBE cells at high doses (≤8 μg/mL). The neutropenic murine thigh infection model study revealed that compound 22c displayed more effective in vivo bactericidal activity than tiamulin in reducing MRSA load. The molecular docking studies indicated that compound 22c was successfully localized inside the binding pocket of 50S ribosomal, and four hydrogen bonds played important roles in the binding of them.

Sonicverse: A Multisensory Simulation Platform for Embodied Household Agents that See and Hear

Abstract: Developing embodied agents in simulation has been a key research topic in recent years. Exciting new tasks, algorithms, and benchmarks have been developed in various simulators. However, most of them assume deaf agents in silent environments, while we humans perceive the world with multiple senses. We introduce Sonicverse, a multisensory simulation platform with integrated audio-visual simulation for training household agents that can both see and hear. Sonicverse models realistic continuous audio rendering in 3D environments in real-time. Together with a new audio-visual VR interface that allows humans to interact with agents with audio, Sonicverse enables a series of embodied AI tasks that need audio-visual perception. For semantic audio-visual navigation in particular, we also propose a new multi-task learning model that achieves state-of-the-art performance. In addition, we demonstrate Sonicverse's realism via sim-to-real transfer, which has not been achieved by other simulators: an agent trained in Sonicverse can successfully perform audio-visual navigation in real-world environments. Sonicverse is available at: https://github.com/StanfordVL/Sonicverse.

Modeling Dynamic Environments with Scene Graph Memory

Abstract: Embodied AI agents that search for objects in large environments such as households often need to make efficient decisions by predicting object locations based on partial information. We pose this as a new type of link prediction problem: link prediction on partially observable dynamic graphs. Our graph is a representation of a scene in which rooms and objects are nodes, and their relationships are encoded in the edges; only parts of the changing graph are known to the agent at each timestep. This partial observability poses a challenge to existing link prediction approaches, which we address. We propose a novel state representation -- Scene Graph Memory (SGM) -- with captures the agent's accumulated set of observations, as well as a neural net architecture called a Node Edge Predictor (NEP) that extracts information from the SGM to search efficiently. We evaluate our method in the Dynamic House Simulator, a new benchmark that creates diverse dynamic graphs following the semantic patterns typically seen at homes, and show that NEP can be trained to predict the locations of objects in a variety of environments with diverse object movement dynamics, outperforming baselines both in terms of new scene adaptability and overall accuracy. The codebase and more can be found at https://www.scenegraphmemory.com.

Nitrogen and phosphorus co-doped activated carbon induces high density Cu+ active center for acetylene hydrochlorination

Abstract: This work aims to solve the problems of low reaction activity of Cu-based catalysts and agglomeration of active centers in acetylene hydrochlorination. Cu-based catalysts supported by N-P co-doped activated carbon (AC) with different content (m Cu-x NP/AC) were manufactured and applied in the acetylene hydrochlorination reaction. It was found that the doping of carriers N and P induced the transformation of Cu2+ to Cu+, and the catalytic activity was markedly improved. Under the optimal reaction temperature of 220 °C, the gas hourly space velocity (GHSV) of C2H2 was 90 h−1 and VHCl:VC2H2 was 1.15. The initial activity of the 5%Cu-30NP/AC catalyst reached 95.59%. Through some characterization methods showed the addition of N and P improved the dispersion of Cu in carbon, which increased the ratio of Cu+/Cu2+. The measurement results confirmed that the chemisorption capacity of m Cu-x NP/AC for C2H2 decreased slightly, and the chemisorption capacity for HCl increased significantly, which was the reason for the increased activity of the catalyst. The conclusion provides a reference for the development of acetylene hydrochlorination Cu catalyst. © 2023 The Chemical Industry and Engineering Society of China, and Chemical Industry Press Co., Ltd.. All rights reserved.

Natural history of isolated abdominal aortic dissection: A prospective cohort study

Abstract: Objectives Isolated abdominal aortic dissection (IAAD) is extremely rare, with its optimal treatment and intervention timing remaining poorly understood. We aimed to study the natural history of IAAD and facilitate better clinical decision. Methods Consecutive patients admitted to our institution from January 2016 to April 2021 were enrolled and followed up prospectively. All-cause death was taken as the primary endpoint. Results A total of 68 patients with IAAD were included. The mean age at presentation was 61.2 ± 14.8 (Range: 26.0, 93.0) years and 55 (80.9%) were male. A total of 38 (55.9%) patients were treated conservatively, 27 (39.7%) received endovascular aneurysm repair (EVAR), and 3 (4.4%) underwent open surgery. After a mean follow-up of 2.4 years (Range: 0.1, 5.5), 9 (13.2%) patients died, 8 of whom (21.0%) were treated conservatively and 1 EVAR (3.7%). Compared with EVAR/open surgery, patient treated conservatively had a much worse survival (p = 0.043). There was no significant difference between different IAAD aortic sizes regarding mortality (p = 0.220). Patients with completely thrombosed false lumen fared improved survival rate, followed by partial thrombosis and patency, respectively, although not significantly (p = 0.190). No significant difference was observed between male and female concerning survival rate (p = 0.970). Patients without symptoms had a significantly improved survival (p = 0.048). Conclusion On the basis of patients’ preference and surgeons’ experience, a more aggressive treatment regimen for IAAD should be considered, with EVAR being the first choice, especially for those with persistent symptoms and patent false lumen, regardless of sex, age, or aortic size.

DPPH Radical Scavenging Activity of New Phenolics from the Fermentation Broth of Mushroom Morehella importuna

Abstract: In recent years, wild morel mushroom species have begun to be widely cultivated in China due to their high edible and medicinal values. To parse the medicinal ingredients, we employed the technique of liquid-submerged fermentation to investigate the secondary metabolites of Morehella importuna. Two new natural isobenzofuranone derivatives (1–2) and one new orsellinaldehyde derivative (3), together with seven known compounds, including one o-orsellinaldehyde (4), phenylacetic acid (5), benzoic acid (6), 4-hydroxy-phenylacetic acid (7), 3,5-dihydroxybenzoic acid (8), N,N′-pentane-1,5-diyldiacetamide (9), and 1H-pyrrole-2-carboxylic acid (10), were obtained from the fermented broth of M. importuna. Their structures were determined according to the data of NMR, HR Q-TOF MS, IR, UV, optical activity, and single-crystal X-ray crystallography. TLC-bioautography displayed that these compounds possess significant antioxidant activity with the half DPPH free radical scavenging concentration of 1.79 (1), 4.10 (2), 4.28 (4), 2.45 (5), 4.40 (7), 1.73 (8), and 6.00 (10) mM. The experimental results would shed light on the medicinal value of M. importuna for its abundant antioxidants.

Editorial: Multimedia environmental pollution and food safety: New insights from integrated consumer nutrition and health risk management

Abstract: the

M-EMBER: Tackling Long-Horizon Mobile Manipulation via Factorized Domain Transfer

Abstract: In this paper, we propose a novel method to create visuomotor mobile manipulation solutions to long-horizon activities. We propose to leverage the recent advances in robot simulation to train robust visual solutions in simulation that can transfer to the real world. While previous works have shown success applying this procedure to autonomous visual navigation and stationary manipulation, applying it to long-horizon visuomotor mobile manipulation is still an open challenge that demands both perceptual and compositional generalization of multiple skills. In this work, we develop Mobile-EMBER, or M-EMBER, a factorized method that decomposes a long-horizon mobile manipulation activity into a repertoire of primitive visual skills, reinforcement-learns each skill in simulation, and composes these skills to a long-horizon mobile manipulation activity. On a real mobile manipulation robot, we find that M-EMBER completes a long-horizon mobile manipulation activity, cleaning_kitchen, achieving over 50% success rate. This requires successfully planning and executing five factorized, learned visual skills, in sequences of up to 48 skills long.

Cloning and functional analysis of RADIALIS-like 1 gene from Antirrhinum majus.

Abstract: MYB is one of the largest transcription factor families in plants. Among them, the R3-MYB transcription factor RADIALIS (RAD) plays a very important role in the flowers development in Antirrhinum majus. In this study, a R3-MYB gene similar to RAD was found by analyzing the genome of A. majus, which was named AmRADIALIS-like 1 (AmRADL1). The gene function was predicted through bioinformatics. The relative expression levels in different tissues and organs of wild-type A. majus were analyzed by qRT-PCR. AmRADL1 was overexpressed in A. majus, and the transgenic plants were analyzed by morphological observation and histological staining. The results showed that the open reading frame (ORF) of AmRADL1 gene was 306 bp in length, encoding 101 amino acids. It has typical SANT domain, and the C-terminal contains a CREB motif, which was highly homologous to tomato SlFSM1. The results of qRT-PCR showed that AmRADL1 was expressed in roots, stems, leaves and flowers, and the expression level was higher in flowers. Further analysis of its expression in different floral organs showed that AmRADL1 had the highest expression in carpel. The results of histological staining analysis of the transgenic plants showed that compared with the wild type, although the size of the carpel cells of the transgenic plants did not change significantly, the placenta area in the carpel became smaller and the number of cell decreased. In summary, AmRADL1 may be involved in the regulation of carpel development, but the specific mechanism of action in carpel remains to be further studied.

Task-Driven Graph Attention for Hierarchical Relational Object Navigation

Abstract: Embodied AI agents in large scenes often need to navigate to find objects. In this work, we study a naturally emerging variant of the object navigation task, hierarchical relational object navigation (HRON), where the goal is to find objects specified by logical predicates organized in a hierarchical structure-objects related to furniture and then to rooms-such as finding an apple on top of a table in the kitchen. Solving such a task requires an efficient representation to reason about object relations and correlate the relations in the environment and in the task goal. HRON in large scenes (e.g. homes) is particularly challenging due to its partial observability and long horizon, which invites solutions that can compactly store the past information while effectively exploring the scene. We demonstrate experimentally that scene graphs are the best-suited representation compared to conventional representations such as images or 2D maps. We propose a solution that uses scene graphs as part of its input and integrates graph neural networks as its backbone, with an integrated task-driven attention mechanism, and demonstrate its better scalability and learning efficiency than state-of-the-art baselines.

VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models

Abstract: Large language models (LLMs) are shown to possess a wealth of actionable knowledge that can be extracted for robot manipulation in the form of reasoning and planning. Despite the progress, most still rely on pre-defined motion primitives to carry out the physical interactions with the environment, which remains a major bottleneck. In this work, we aim to synthesize robot trajectories, i.e., a dense sequence of 6-DoF end-effector waypoints, for a large variety of manipulation tasks given an open-set of instructions and an open-set of objects. We achieve this by first observing that LLMs excel at inferring affordances and constraints given a free-form language instruction. More importantly, by leveraging their code-writing capabilities, they can interact with a visual-language model (VLM) to compose 3D value maps to ground the knowledge into the observation space of the agent. The composed value maps are then used in a model-based planning framework to zero-shot synthesize closed-loop robot trajectories with robustness to dynamic perturbations. We further demonstrate how the proposed framework can benefit from online experiences by efficiently learning a dynamics model for scenes that involve contact-rich interactions. We present a large-scale study of the proposed method in both simulated and real-robot environments, showcasing the ability to perform a large variety of everyday manipulation tasks specified in free-form natural language. Project website: https://voxposer.github.io

Otomastoiditis Caused by Nocardia Farcinica: A Case Report and Literature Review.

Abstract: Nocardia farcinica usually infects people with impaired immune status and usually manifests in the lungs. Otomastoiditis caused by Nocardia infection is extremely rare, with only 4 cases reported to date. This report describes a case of otomastoid N. farcinica infection in an immunocompetent patient. The case was a 10-month-old immunocompetent infant who presented with an approximately 3-month history of right ear discharge for which treatment with various antibiotics had not resulted in significant improvement. Multiple cultures of secretions and pathologic examination failed to identify the causative organism. The patient then underwent right mastoidectomy. Finally, metagenomic next-generation sequencing identified the pathogen to be N. farcinica. The patient was infection-free at the 6-month follow-up but had developed labyrinthitis ossificans. Otomastoid Nocardia infection has characteristic clinical features, namely, formation of a large amount of granulation tissue and coexistence of bone destruction and new bone formation. Traditionally, Nocardia is challenging to diagnose. Metagenomic next-generation sequencing of lesions is helpful. Complete local debridement and free drainage are key to treatment.

Robust Model Predictive Current Control With Adaptive Extended-State-Observer for Permanent Magnet Synchronous Motor

Abstract: Three-vector based model predictive current control (TV-MPCC) has been recently proposed to improve the steady-state performance for permanent magnet synchronous motor(PMSM) drive system. However, the TV-MPCC presents poor performance when the predictive model is inaccurate due to motor parameter mismatches. To address this issue, a TVMPCC based on adaptive extended-state observer (TV-MPCCAESO) for PMSM is proposed in this paper. Therein, the disturbance caused by parameter mismatches is observed by the designed AESO and compensated into the predictive model to improve the robustness of TV-MPCC. Extensive comparative simulations are conducted to demonstrate the effectiveness of the proposed TV-MPCC-AESO scheme.

Differentially Private Video Activity Recognition

Abstract: In recent years, differential privacy has seen significant advancements in image classification; however, its application to video activity recognition remains under-explored. This paper addresses the challenges of applying differential privacy to video activity recognition, which primarily stem from: (1) a discrepancy between the desired privacy level for entire videos and the nature of input data processed by contemporary video architectures, which are typically short, segmented clips; and (2) the complexity and sheer size of video datasets relative to those in image classification, which render traditional differential privacy methods inadequate. To tackle these issues, we propose Multi-Clip DP-SGD, a novel framework for enforcing video-level differential privacy through clip-based classification models. This method samples multiple clips from each video, averages their gradients, and applies gradient clipping in DP-SGD without incurring additional privacy loss. Moreover, we incorporate a parameter-efficient transfer learning strategy to make the model scalable for large-scale video datasets. Through extensive evaluations on the UCF-101 and HMDB-51 datasets, our approach exhibits impressive performance, achieving 81% accuracy with a privacy budget of epsilon=5 on UCF-101, marking a 76% improvement compared to a direct application of DP-SGD. Furthermore, we demonstrate that our transfer learning strategy is versatile and can enhance differentially private image classification across an array of datasets including CheXpert, ImageNet, CIFAR-10, and CIFAR-100.

Dynamic-Resolution Model Learning for Object Pile Manipulation

Abstract: Dynamics models learned from visual observations have shown to be effective in various robotic manipulation tasks. One of the key questions for learning such dynamics models is what scene representation to use. Prior works typically assume representation at a fixed dimension or resolution, which may be inefficient for simple tasks and ineffective for more complicated tasks. In this work, we investigate how to learn dynamic and adaptive representations at different levels of abstraction to achieve the optimal trade-off between efficiency and effectiveness. Specifically, we construct dynamic-resolution particle representations of the environment and learn a unified dynamics model using graph neural networks (GNNs) that allows continuous selection of the abstraction level. During test time, the agent can adaptively determine the optimal resolution at each model-predictive control (MPC) step. We evaluate our method in object pile manipulation, a task we commonly encounter in cooking, agriculture, manufacturing, and pharmaceutical applications. Through comprehensive evaluations both in the simulation and the real world, we show that our method achieves significantly better performance than state-of-the-art fixed-resolution baselines at the gathering, sorting, and redistribution of granular object piles made with various instances like coffee beans, almonds, corn, etc.

Investigation and Analysis of Private Doctors Helping the Elderly Care Market

Abstract: China has a high degree of population aging, pension has become a hot topic. Based on this, this study focuses on the private doctor service assisted pension market, and investigates doctors, young people and the elderly respectively. The correlation analysis was used to obtain the relationship between respondents' willingness to accept private doctors and various factors. Further with the help of Logistic regression model, it is concluded that doctors in private hospitals, young and middle-aged elderly men and elderly people living alone and with high cost of living are more willing to accept the service of private doctors.

Preparation of Red Iron by Magnetization Roasting-Hydrothermal Method Using Ultra-Low-Grade Limonite

Abstract: Iron is one of the most important strategic materials in national production, and the demand for iron ore is huge in the world. High quality iron ore reserves have been almost exhausted, and it is necessary to develop a technology that utilizes low-grade iron ore. Limonite is a representative low-grade iron ore due to its complex mineral and elemental composition. In this paper, the union process was employed to separate the iron elements in low-grade limonite. Firstly, a rough iron concentrate was obtained under 1.0 T of magnetic field intensity and −0.074 mm > 94.84% of grinding fineness; then, the rough iron concentrate was magnetization roasted under a temperature of 700 °C, 60 min of retention time, 3 wt% of biochar consumption, and 0.15 T of magnetic field intensity. The grade of iron concentrate was 59.57% and the recovery of iron was 90.72%. Finally, the red iron pigment was produced via a high temperature hydrothermal method in order to increase the additional value of this ultra-low-grade limonite. The optimal parameters were 10.0 g/L of solution acidity, a 200 °C reaction temperature, 5 h of reaction time, and a 6:1 solid-to-liquid ratio. The reaction mechanism was also discussed.

HomE: Homography-Equivariant Video Representation Learning

Abstract: Recent advances in self-supervised representation learning have enabled more efficient and robust model performance without relying on extensive labeled data. However, most works are still focused on images, with few working on videos and even fewer on multi-view videos, where more powerful inductive biases can be leveraged for self-supervision. In this work, we propose a novel method for representation learning of multi-view videos, where we explicitly model the representation space to maintain Homography Equivariance (HomE). Our method learns an implicit mapping between different views, culminating in a representation space that maintains the homography relationship between neighboring views. We evaluate our HomE representation via action recognition and pedestrian intent prediction as downstream tasks. On action classification, our method obtains 96.4% 3-fold accuracy on the UCF101 dataset, better than most state-of-the-art self-supervised learning methods. Similarly, on the STIP dataset, we outperform the state-of-the-art by 6% for pedestrian intent prediction one second into the future while also obtaining an accuracy of 91.2% for pedestrian action (cross vs. not-cross) classification. Code is available at https://github.com/anirudhs123/HomE.

2022 CMICS Expert Consensus on the Management of Isolated Tricuspid Regurgitation after Left-Sided Valve Surgery

Abstract: Tricuspid regurgitation (TR) may occur late after left-sided valve surgery (LSVS). Isolated tricuspid regurgitation after left-sided valve surgery (iTR-LSVS) refers to isolated TR without significant lesions in the mitral and/or aortic position late after mitral and/or aortic replacement or repair. Severe TR has a negative impact on long-term prognosis and requires surgical or transcatheter treatment. However, there is no clear recommendation on when and how intervention should be performed for patients with iTR-LSVS in the current guidelines for the management of valvular heart disease. The historically high operative mortality may be reduced by current minimally invasive techniques and transcatheter therapy. To further understand iTR-LSVS, standardize the treatment, improve the prognosis, and promote the collaboration, the Chinese Minimally Invasive Cardiovascular Surgery Committee (CMICS) wrote this expert consensus on the management of iTR-LSVS from the aspects of etiology, preoperative evaluation, indications for intervention, surgical treatment, transcatheter therapy, and postoperative management.