Soft Robotic Finger with Variable Effective Length enabled by an Antagonistic Constraint Mechanism

Compared to traditional rigid robotics, soft robotics has attracted increasing attention due to its advantages as compliance, safety, and low cost. As an essential part of soft robotics, the soft robotic gripper also shows its superior while grasping the objects with irregular shapes. Recent research has been conducted to improve its grasping performance by adjusting the variable effective length (VEL). However, the VEL achieved by multi-chamber design or tunable stiffness shape memory material requires complex pneumatic circuit design or a time-consuming phase-changing process. This work proposes a fold-based soft robotic actuator made from 3D printed filament, NinjaFlex. It is experimentally tested and represented by the hyperelastic model. Mathematic and finite element modelling is conducted to study the bending behaviour of the proposed soft actuator. Besides, an antagonistic constraint mechanism is proposed to achieve the VEL, and the experiments demonstrate that better conformity is achieved. Finally, a two-mode gripper is designed and evaluated to demonstrate the advances of VEL on grasping performance.

Geometry-Aware Fruit Grasping Estimation for Robotic Harvesting in Orchards

Field robotic harvesting is a promising technique in recent development of agricultural industry. It is vital for robots to recognise and localise fruits before the harvesting in natural orchards. However, the workspace of harvesting robots in orchards is complex: many fruits are occluded by branches and leaves. It is important to estimate a proper grasping pose for each fruit before performing the manipulation. In this study, a geometry-aware network, A3N, is proposed to perform end-to-end instance segmentation and grasping estimation using both color and geometry sensory data from a RGB-D camera. Besides, workspace geometry modelling is applied to assist the robotic manipulation. Moreover, we implement a global-to-local scanning strategy, which enables robots to accurately recognise and retrieve fruits in field environments with two consumer-level RGB-D cameras. We also evaluate the accuracy and robustness of proposed network comprehensively in experiments. The experimental results show that A3N achieves 0.873 on instance segmentation accuracy, with an average computation time of 35 ms. The average accuracy of grasping estimation is 0.61 cm and 4.8$^{\circ}$ in centre and orientation, respectively. Overall, the robotic system that utilizes the global-to-local scanning and A3N, achieves success rate of harvesting ranging from 70\% - 85\% in field harvesting experiments.

Adaptive Multi-receptive Field Spatial-Temporal Graph Convolutional Network for Traffic Forecasting

Mobile network traffic forecasting is one of the key functions in daily network operation. A commercial mobile network is large, heterogeneous, complex and dynamic. These intrinsic features make mobile network traffic forecasting far from being solved even with recent advanced algorithms such as graph convolutional network-based prediction approaches and various attention mechanisms, which have been proved successful in vehicle traffic forecasting. In this paper, we cast the problem as a spatial-temporal sequence prediction task. We propose a novel deep learning network architecture, Adaptive Multi-receptive Field Spatial-Temporal Graph Convolutional Networks (AMF-STGCN), to model the traffic dynamics of mobile base stations. AMF-STGCN extends GCN by (1) jointly modeling the complex spatial-temporal dependencies in mobile networks, (2) applying attention mechanisms to capture various Receptive Fields of heterogeneous base stations, and (3) introducing an extra decoder based on a fully connected deep network to conquer the error propagation challenge with multi-step forecasting. Experiments on four real-world datasets from two different domains consistently show AMF-STGCN outperforms the state-of-the-art methods.

Failure-averse Active Learning for Physics-constrained Systems

Active learning is a subfield of machine learning that is devised for design and modeling of systems with highly expensive sampling costs. Industrial and engineering systems are generally subject to physics constraints that may induce fatal failures when they are violated, while such constraints are frequently underestimated in active learning. In this paper, we develop a novel active learning method that avoids failures considering implicit physics constraints that govern the system. The proposed approach is driven by two tasks: the safe variance reduction explores the safe region to reduce the variance of the target model, and the safe region expansion aims to extend the explorable region exploiting the probabilistic model of constraints. The global acquisition function is devised to judiciously optimize acquisition functions of two tasks, and its theoretical properties are provided. The proposed method is applied to the composite fuselage assembly process with consideration of material failure using the Tsai-wu criterion, and it is able to achieve zero-failure without the knowledge of explicit failure regions.

A Tactile-enabled Grasping Method for Robotic Fruit Harvesting

In the robotic crop harvesting environment, foreign objects intrusion in the gripper workspace is frequently occurring and unignorable, however, rarely addressed. This paper presents a novel intelligent robotic grasping method capable of handling obstacle interference, which is the first of its kind in the literature. The proposed method combines deep learning algorithms with low-cost tactile sensing hardware on a multi-DoF soft robotic gripper. Through experimental validations, the proposed method demonstrated promising performance in distinguishing various grasping scenarios. The 4-finger independently controlled gripper presented outstanding adaptability to handle various picking scenarios. The overall performance of this work indicated great potential for solving the robotic fruit harvesting challenges.

Self-Training Sampling with Monolingual Data Uncertainty for Neural Machine Translation

Self-training has proven effective for improving NMT performance by augmenting model training with synthetic parallel data. The common practice is to construct synthetic data based on a randomly sampled subset of large-scale monolingual data, which we empirically show is sub-optimal. In this work, we propose to improve the sampling procedure by selecting the most informative monolingual sentences to complement the parallel data. To this end, we compute the uncertainty of monolingual sentences using the bilingual dictionary extracted from the parallel data. Intuitively, monolingual sentences with lower uncertainty generally correspond to easy-to-translate patterns which may not provide additional gains. Accordingly, we design an uncertainty-based sampling strategy to efficiently exploit the monolingual data for self-training, in which monolingual sentences with higher uncertainty would be sampled with higher probability. Experimental results on large-scale WMT English$\Rightarrow$German and English$\Rightarrow$Chinese datasets demonstrate the effectiveness of the proposed approach. Extensive analyses suggest that emphasizing the learning on uncertain monolingual sentences by our approach does improve the translation quality of high-uncertainty sentences and also benefits the prediction of low-frequency words at the target side.

TranSmart: A Practical Interactive Machine Translation System

Automatic machine translation is super efficient to produce translations yet their quality is not guaranteed. This technique report introduces TranSmart, a practical human-machine interactive translation system that is able to trade off translation quality and efficiency. Compared to existing publicly available interactive translation systems, TranSmart supports three key features, word-level autocompletion, sentence-level autocompletion and translation memory. By word-level and sentence-level autocompletion, TranSmart allows users to interactively translate words in their own manners rather than the strict manner from left to right. In addition, TranSmart has the potential to avoid similar translation mistakes by using translated sentences in history as its memory. This report presents major functions of TranSmart, algorithms for achieving these functions, how to use the TranSmart APIs, and evaluation results of some key functions. TranSmart is publicly available at its homepage (https://transmart.qq.com).

Benchmarking Graph Neural Networks on Link Prediction

In this paper, we benchmark several existing graph neural network (GNN) models on different datasets for link predictions. In particular, the graph convolutional network (GCN), GraphSAGE, graph attention network (GAT) as well as variational graph auto-encoder (VGAE) are implemented dedicated to link prediction tasks, in-depth analysis are performed, and results from several different papers are replicated, also a more fair and systematic comparison are provided. Our experiments show these GNN architectures perform similarly on various benchmarks for link prediction tasks.

Hierarchical Representation via Message Propagation for Robust Model Fitting

In this paper, we propose a novel hierarchical representation via message propagation (HRMP) method for robust model fitting, which simultaneously takes advantages of both the consensus analysis and the preference analysis to estimate the parameters of multiple model instances from data corrupted by outliers, for robust model fitting. Instead of analyzing the information of each data point or each model hypothesis independently, we formulate the consensus information and the preference information as a hierarchical representation to alleviate the sensitivity to gross outliers. Specifically, we firstly construct a hierarchical representation, which consists of a model hypothesis layer and a data point layer. The model hypothesis layer is used to remove insignificant model hypotheses and the data point layer is used to remove gross outliers. Then, based on the hierarchical representation, we propose an effective hierarchical message propagation (HMP) algorithm and an improved affinity propagation (IAP) algorithm to prune insignificant vertices and cluster the remaining data points, respectively. The proposed HRMP can not only accurately estimate the number and parameters of multiple model instances, but also handle multi-structural data contaminated with a large number of outliers. Experimental results on both synthetic data and real images show that the proposed HRMP significantly outperforms several state-of-the-art model fitting methods in terms of fitting accuracy and speed.

Multi-Task Learning with Shared Encoder for Non-Autoregressive Machine Translation

Non-Autoregressive machine Translation (NAT) models have demonstrated significant inference speedup but suffer from inferior translation accuracy. The common practice to tackle the problem is transferring the Autoregressive machine Translation (AT) knowledge to NAT models, e.g., with knowledge distillation. In this work, we hypothesize and empirically verify that AT and NAT encoders capture different linguistic properties and representations of source sentences. Therefore, we propose to adopt the multi-task learning to transfer the AT knowledge to NAT models through the encoder sharing. Specifically, we take the AT model as an auxiliary task to enhance NAT model performance. Experimental results on WMT14 English->German and WMT16 English->Romanian datasets show that the proposed multi-task NAT achieves significant improvements over the baseline NAT models. In addition, experimental results demonstrate that our multi-task NAT is complementary to the standard knowledge transfer method, knowledge distillation. Code is publicly available at https://github.com/yongchanghao/multi-task-nat