Objects In Clutter Research Articles

PolyDexFrame: Deep Reinforcement Learning-Based Pick-and-Place of Objects in Clutter

This research study represents a polydexterous deep reinforcement learning-based pick-and-place framework for industrial clutter scenarios. In the proposed framework, the agent tends to learn the pick-and-place of regularly and irregularly shaped objects in clutter by using the sequential combination of prehensile and non-prehensile robotic manipulations involving different robotic grippers in a completely self-supervised manner. The problem was tackled as a reinforcement learning problem; after the Markov decision process (MDP) was designed, the off-policy model-free Q-learning algorithm was deployed using deep Q-networks as a Q-function approximator. Four distinct robotic manipulations, i.e., grasp from the prehensile manipulation category and inward slide, outward slide, and suction grip from the non-prehensile manipulation category were considered as actions. The Q-function comprised four fully convolutional networks (FCN) corresponding to each action based on memory-efficient DenseNet-121 variants outputting pixel-wise maps of action-values jointly trained via the pixel-wise parametrization technique. Rewards were awarded according to the status of the action performed, and backpropagation was conducted accordingly for the FCN generating the maximum Q-value. The results showed that the agent learned the sequential combination of the polydexterous prehensile and non-prehensile manipulations, where the non-prehensile manipulations increased the possibility of prehensile manipulations. We achieved promising results in comparison to the baselines, differently designed variants, and density-based testing clutter.

Efficient push-grasping for multiple target objects in clutter environments.

Intelligent manipulation of robots in an unstructured environment is an important application field of artificial intelligence, which means that robots must have the ability of autonomous cognition and decision-making. A typical example of this type of environment is a cluttered scene where objects are stacked and close together. In clutter, the target(s) may be one or more, and efficiently completing the target(s) grasping task is challenging. In this study, an efficient push-grasping method based on reinforcement learning is proposed for multiple target objects in clutter. The key point of this method is to consider the states of all the targets so that the pushing action can expand the grasping space of all targets as much as possible to achieve the minimum total number of pushing and grasping actions and then improve the efficiency of the whole system. At this point, we adopted the mask fusion of multiple targets, clearly defined the concept of graspable probability, and provided the reward mechanism of multi-target push-grasping. Experiments were conducted in both the simulation and real systems. The experimental results indicated that, compared with other methods, the proposed method performed better for multiple target objects and a single target in clutter. It is worth noting that our policy was only trained under simulation, which was then transferred to the real system without retraining or fine-tuning.

Effects of involuntary and voluntary attention on critical spacing of visual crowding.

Visual spatial attention can be allocated in two distinct ways: one that is voluntarily directed to behaviorally relevant locations in the world, and one that is involuntarily captured by salient external stimuli. Precueing spatial attention has been shown to improve perceptual performance on a number of visual tasks. However, the effects of spatial attention on visual crowding, defined as the reduction in the ability to identify target objects in clutter, are far less clear. In this study, we used an anticueing paradigm to separately measure the effects of involuntary and voluntary spatial attention on a crowding task. Each trial began with a brief peripheral cue that predicted that the crowded target would appear on the opposite side of the screen 80% of the time and on the same side of the screen 20% of the time. Subjects performed an orientation discrimination task on a target Gabor patch that was flanked by other similar Gabor patches with independent random orientations. For trials with a short stimulus onset asynchrony between cue and target, involuntary capture of attention led to faster response times and smaller critical spacing when the target appeared on the cue side. For trials with a long stimulus onset asynchrony, voluntary allocation of attention led to faster reaction times but no significant effect on critical spacing when the target appeared on the opposite side to the cue. We additionally found that the magnitudes of these cueing effects of involuntary and voluntary attention were not strongly correlated across subjects for either reaction time or critical spacing.

A Passively Conforming Soft Robotic Gripper with Three-Dimensional Negative Bending Stiffness Fingers.

Robot grippers that lack physical compliance have a difficult time dealing with uncertainty, such as fragile objects that may not have well-defined shapes. Existing soft robotic grippers require a large empty workspace for their actuated fingers to curl around the objects of interest, limiting their performance in clutter. This article presents a three-dimensional structure that exhibits negative stiffness in every bending direction used as fingers in a class of soft robotic grippers. Our approach exploits a compliant mechanism in a conical shape such that a transverse external contact force causes the fingers to bend toward the contact, enabling passive conformation for an adaptive grasp, even in clutter. We show analytically and experimentally that the proposed fingers have a negative bending response and that they conform to objects of various diameters. We demonstrate a soft robotic gripper with three self-conforming fingers performing the following: (1) fingertip grasping, (2) power grasping, and (3) semipassive grasping in clutter. Grasping experiments focus on picking fruits, which exemplify delicate objects with unmodeled shapes with significant variation. The experimental results reveal the ability of the self-conforming structure to smoothly envelope a broad range of objects and demonstrate a 100% grasp success rate in the experiments performed. The proposed passively conforming fingers enable picking of complex and unknown geometries without disturbing nearby objects in clutter and without the need for complex grasping algorithms. The proposed structures can be tailored to deform in desired ways, enabling a robust strategy for the engineering of physical compliance for adaptive soft structures.

Prehensile and Non-Prehensile Robotic Pick-and-Place of Objects in Clutter Using Deep Reinforcement Learning.

In this study, we develop a framework for an intelligent and self-supervised industrial pick-and-place operation for cluttered environments. Our target is to have the agent learn to perform prehensile and non-prehensile robotic manipulations to improve the efficiency and throughput of the pick-and-place task. To achieve this target, we specify the problem as a Markov decision process (MDP) and deploy a deep reinforcement learning (RL) temporal difference model-free algorithm known as the deep Q-network (DQN). We consider three actions in our MDP; one is 'grasping' from the prehensile manipulation category and the other two are 'left-slide' and 'right-slide' from the non-prehensile manipulation category. Our DQN is composed of three fully convolutional networks (FCN) based on the memory-efficient architecture of DenseNet-121 which are trained together without causing any bottleneck situations. Each FCN corresponds to each discrete action and outputs a pixel-wise map of affordances for the relevant action. Rewards are allocated after every forward pass and backpropagation is carried out for weight tuning in the corresponding FCN. In this manner, non-prehensile manipulations are learnt which can, in turn, lead to possible successful prehensile manipulations in the near future and vice versa, thus increasing the efficiency and throughput of the pick-and-place task. The Results section shows performance comparisons of our approach to a baseline deep learning approach and a ResNet architecture-based approach, along with very promising test results at varying clutter densities across a range of complex scenario test cases.

FESNet: Frequency-Enhanced Saliency Detection Network for Grain Pest Segmentation

As insect infestation is the leading factor accounting for nutritive and economic losses in stored grains, it is important to detect the presence and number of insects for the sake of taking proper control measures. Inspired by the human visual attention mechanism, we propose a U-net-like frequency-enhanced saliency (FESNet) detection model, resulting in the pixelwise segmentation of grain pests. The frequency clues, as well as the spatial information, are leveraged to enhance the detection performance of small insects from the cluttered grain background. Firstly, we collect a dedicated dataset, GrainPest, with pixel-level annotation after analyzing the image attributes of the existing salient object detection datasets. Secondly, we design a FESNet with the discrete wavelet transformation (DWT) and the discrete cosine transformation (DCT), both involved in the traditional convolution layers. As current salient object detection models will reduce the spatial information with pooling operations in the sequence of encoding stages, a special branch of the discrete wavelet transformation (DWT) is connected to the higher stages to capture accurate spatial information for saliency detection. Then, we introduce the discrete cosine transform (DCT) into the backbone bottlenecks to enhance the channel attention with low-frequency information. Moreover, we also propose a new receptive field block (NRFB) to enlarge the receptive fields by aggregating three atrous convolution features. Finally, in the phase of decoding, we use the high-frequency information and aggregated features together to restore the saliency map. Extensive experiments and ablation studies on our dataset, GrainPest, and open dataset, Salient Objects in Clutter (SOC), demonstrate that the proposed model performs favorably against the state-of-the-art model.

Few-Shot Instance Grasping of Novel Objects in Clutter

Instance grasping, which aims to grasp a specific object out of clutter, is a fundamental task within robotics. However, allowing a robot to quickly learn to perform instance grasping for new, previously unseen objects remains challenging. In this work, we present an instance grasping meta-learning framework (IGML), a simple yet effective end-to-end approach that not only teaches robots to identify novel objects but also how to grasp them. Given only a few examples to specify the grasping point of the target object, our IGML can quickly learn to recognize the target object and grasp it at the demonstrated grasping point by leveraging prior experience. Experimental results on the test sets show that IGML achieves decent success rates in cluttered environments, significantly surpassing state-of-the-art methods. Then we deployed IGML on a UR5 robot arm to handle pick-and-place scenarios and achieved a precision rate of 93.4% and a recall rate of 87.1%.

Mixture-modeling approach reveals global and local processes in visual crowding

Crowding refers to the inability to recognize objects in clutter, setting a fundamental limit on various perceptual tasks such as reading and facial recognition. While prevailing models suggest that crowding is a unitary phenomenon occurring at an early level of processing, recent studies have shown that crowding might also occur at higher levels of representation. Here we investigated whether local and global crowding interference co-occurs within the same display. To do so, we tested the distinctive contribution of local flanker features and global configurations of the flankers on the pattern of crowding errors. Observers (n = 27) estimated the orientation of a target when presented alone or surrounded by flankers. Flankers were grouped into a global configuration, forming an illusory rectangle when aligned or a rectangular configuration when misaligned. We analyzed the error distributions by fitting probabilistic mixture models. Results showed that participants often misreported the orientation of a flanker instead of that of the target. Interestingly, in some trials the orientation of the global configuration was misreported. These results suggest that crowding occurs simultaneously across multiple levels of visual processing and crucially depends on the spatial configuration of the stimulus. Our results pose a challenge to models of crowding with an early single pooling stage and might be better explained by models which incorporate the possibility of multilevel crowding and account for complex target-flanker interactions.

Excavation Reinforcement Learning Using Geometric Representation

Excavation of irregular rigid objects in clutter, such as fragmented rocks and wood blocks, is very challenging due to their complex interaction dynamics and highly variable geometries. In this paper, we adopt reinforcement learning (RL) to tackle this challenge and learn policies to plan for a sequence of excavation trajectories for irregular rigid objects, given point clouds of excavation scenes.Moreover, we separately learn a compact representation of the point cloud on geometric tasks that do not require human labeling. We show that using the representation reduces training time for RL, while achieving similar asymptotic performance compare to an end-to-end RL algorithm. When using a policy trained in simulation directly on a real scene, we show that the policy trained with the representation outperforms end-to-end RL. To our best knowledge, this letter presents the first application of RL to plan a sequence of excavation trajectories of irregular rigid objects in clutter.

A CNN-Based Fast Picking Method for WEEE Recycling

Recycling of Waste Electrical and Electronic Equipment (WEEE) is challenging due to the high variety in both the design and composition of the products in this waste stream. Therefore, targeted demanufacturing and appropriate material sorting processes are crucial, but remain often a manual and thus labor-intensive and costly activity. Despite first steps towards the automation of these sorting processes can be noticed, many of them are in material sorting, but very few in the sorting at product level. In this perspective, one of the key tasks is the automated classification, picking and manipulation of WEEE products.In the presented research, a closed-loop grasp planning method is proposed for the random picking of WEEE products. The presented grasping method combines a Convolutional Neural Network-based quality prediction and Closed-Loop control, named CNNB-CL, to find the optimized grasp region for unknown objects in a dense clutter. A large-scale dataset is generated for the CNN training, containing over 2.3 million synthetic grasps, and their corresponding grasp qualities evaluated by grasp simulations using 3D models. The proposed 7-layer light-weighted CNN quantitatively estimates the grasp quality of each grasp candidate. In addition, the CNNB-CL utilizes the feedback from a force-torque sensor to detect the optimized grasp and to adjust the grasp strategy. Various empirical experiments prove that the CNNB-CL method finds an optimized grasp for a WEEE clutter from 2,000 grasp candidates within 0.3 s and achieves an average success rate of 92% for different WEEE clutters. Furthermore, the average picking speed is as high as 1,400 mm/s for the random picking of small WEEE products (< 300 g) with a FANUC Delta robot (M-2i A 3SL). The fast-picking ability of the presented method reveals its potential application for the (pre-) processing of WEEE.

Real-time pixel-wise grasp affordance prediction based on multi-scale context information fusion

PurposeThis paper aims to use fully convolutional network (FCN) to predict pixel-wise antipodal grasp affordances for unknown objects and improve the grasp detection performance through multi-scale feature fusion.Design/methodology/approachA modified FCN network is used as the backbone to extract pixel-wise features from the input image, which are further fused with multi-scale context information gathered by a three-level pyramid pooling module to make more robust predictions. Based on the proposed unify feature embedding framework, two head networks are designed to implement different grasp rotation prediction strategies (regression and classification), and their performances are evaluated and compared with a defined point metric. The regression network is further extended to predict the grasp rectangles for comparisons with previous methods and real-world robotic grasping of unknown objects.FindingsThe ablation study of the pyramid pooling module shows that the multi-scale information fusion significantly improves the model performance. The regression approach outperforms the classification approach based on same feature embedding framework on two data sets. The regression network achieves a state-of-the-art accuracy (up to 98.9%) and speed (4 ms per image) and high success rate (97% for household objects, 94.4% for adversarial objects and 95.3% for objects in clutter) in the unknown object grasping experiment.Originality/valueA novel pixel-wise grasp affordance prediction network based on multi-scale feature fusion is proposed to improve the grasp detection performance. Two prediction approaches are formulated and compared based on the proposed framework. The proposed method achieves excellent performances on three benchmark data sets and real-world robotic grasping experiment.

Planning for grasping cluttered objects based on obstruction degree

Grasping objects in clutter is more difficult than grasping a separated single object. An important issue is that unsafe grasps may occur, in case, one object sits or leans on another, which could cause the collapse of objects. In addition, reachability of each object surrounded by other obstacles also has to be considered. So the order of multiple objects for grasping and the grasp configuration of each object must be planned simultaneously. This article combines grasp order and grasp configuration planning to perform fast and safe multiobject grasping in cluttered scenes. First, a comprehensive grasp configuration database is built to provide enough feasible grasp configurations for the objects. Then, we propose an obstruction degree to estimate the likelihood of reachability of each grasp configuration as well as each object. This measurement also implicitly infers object interactions. Finally, grasp order and grasp configurations are planned together to deal with the constraints caused by reachability and object interaction. Simulations and experiments in a series of cluttered scenes demonstrate that our method can grasp objects efficiently and can greatly reduce unsafe grasps.

Excavation Learning for Rigid Objects in Clutter

Autonomous excavation for hard or compact materials, especially irregular rigid objects, is challenging due to high variance of geometric and physical properties of objects, and large resistive force during excavation. In this letter, we propose a novel learning-based excavation planning method for rigid objects in clutter. Our method consists of a convolutional neural network to predict the excavation success and a sampling-based optimization method for planning high-quality excavation trajectories leveraging the learned prediction model. To reduce the sim2real gap for excavation learning, we propose a voxel-based representation of the excavation scene. We perform excavation experiments in both simulation and real world to evaluate the learning-based excavation planners. We further compare with two heuristic baseline excavation planners and one data-driven scene-independent planner. The experimental results show that our method can plan high-quality excavations for rigid objects in clutter and outperforms the baseline methods by large margins. As far as we know, our work presents the first learning-based excavation planner for cluttered and irregular rigid objects.

Direction-Based Hybrid Strategy Combining Pushing and Hitting for Fast Object Singulation

This paper presents a hybrid singulation strategy for fast object singulation in a cluttered environment. Recent techniques related to object singulation in clutter have employed various kinds of pushing techniques and in some cases have also used hitting techniques. However, these techniques have not addressed the issue related to the direction of pushing and hitting which is vital for fast object singulation. Finding the appropriate direction of hitting and pushing helps in singulating objects quickly in a cluttered environment. This paper proposes the desired direction for pushing and hitting, combined with a hybrid strategy, that results in fast object singulation in a cluttered environment. The number of times of pushing and hitting in terms of time is chosen as the measure of performance. We employ multiple circular disks as the test example and carry out diverse experiments to corroborate the usefulness of the proposed object singulation algorithm. This approach is able to singulate objects quickly in complex formations. In this paper, we have combined both pushing and hitting and also proposed the direction of hitting and pushing in order to singulate objects in clutter quickly.

Unreal mask: one-shot multi-object class-based pose estimation for robotic manipulation using keypoints with a synthetic dataset

Object pose estimation is a prerequisite for many robotic applications. Preparing dataset for network training is a challenging part of the pose estimation approaches, and in most of them, the network can detect just the trained objects. Synthetic data are used to train deep neural networks in robotic manipulation as a promising method for obtaining a huge amount of prelabeled training data, which are generated safely. We are to investigate the reality gap in the pose estimation of intra-category objects from a single RGB-D image using keypoints. The proposed approach in this paper provides a fast and simple procedure for training a deep neural network to identify the object and its keypoints based on synthetic dataset and autolabeling program. To our knowledge, this is the first deep network trained only on synthetic data that can find keypoints of intra-category objects for pose estimation purposes. The speed of training and the simplicity of this method make it very easy to add a new class of objects to the system which is the main advantage of this approach. Using this approach, we demonstrate a near-real-time system estimating object poses with sufficient accuracy for real-world semantic grasping and manipulating of intra-category objects in clutter by a real robot.

Performance Study of Distance-Weighting Approach with Loopy Sum-Product Algorithm for Multi-Object Tracking in Clutter.

In this paper, we explore the performance of the distance-weighting probabilistic data association (DWPDA) approach in conjunction with the loopy sum-product algorithm (LSPA) for tracking multiple objects in clutter. First, we discuss the problem of data association (DA), which is to infer the correspondence between targets and measurements. DA plays an important role when tracking multiple targets using measurements of uncertain origin. Second, we describe three methods of data association: probabilistic data association (PDA), joint probabilistic data association (JPDA), and LSPA. We then apply these three DA methods for tracking multiple crossing targets in cluttered environments, e.g., radar detection with false alarms and missed detections. We are interested in two performance metrics: tracking accuracy and computation time. LSPA is known to be superior to PDA in terms of the former and to dominate JPDA in terms of the latter. Last, we consider an additional DA method that is a modification of PDA by incorporating a weighting scheme based on distances between position estimates and measurements. This distance-weighting approach, when combined with PDA, has been shown to enhance the tracking accuracy of PDA without significant change in the computation burden. Since PDA constitutes a crucial building block of LSPA, we hypothesize that DWPDA, when integrated with LSPA, would perform better under the two performance metrics above. Contrary to expectations, the distance-weighting approach does not enhance the performance of LSPA, whether in terms of tracking accuracy or computation time.

Clustering leads to underestimation of numerosity, but crowding is not the cause

Humans have the remarkable ability to rapidly estimate the number of objects in a visual scene without relying on counting, something referred to as the number sense. It has been well documented that the more clustered the elements are, the lower their perceived numerosity is. A recent account of this observation is the crowdinghypothesis, which posits that the perceived underestimation is driven by visual crowding: the inability to recognise objects in clutter. Crowding can impair individuation of the elements, which would explain the underestimation. Here, we tested the crowding hypothesis by assessing numerosity estimation and crowding for the same stimulus configurations in the same participants. Experiment 1 compared the two tasks when numerosity can be considered to be estimated directly by the visual system (reference patch density = 0.12 items/deg2), while Experiment 2 used high density stimuli (density = 0.88 items/deg2), where numerosity may be estimated indirectly. In both cases, we found that spacing and similarity between elements affected estimation and crowding tasks in markedly different ways. These results are incompatible with a crowding account of numerosity underestimation and point to separate mechanisms for object identification and number estimation, although grouping may play a moderating role in both cases.

Adversarial discriminative sim-to-real transfer of visuo-motor policies

Various approaches have been proposed to learn visuo-motor policies for real-world robotic applications. One solution is first learning in simulation then transferring to the real world. In the transfer, most existing approaches need real-world images with labels. However, the labeling process is often expensive or even impractical in many robotic applications. In this article, we introduce an adversarial discriminative sim-to-real transfer approach to reduce the amount of labeled real data required. The effectiveness of the approach is demonstrated with modular networks in a table-top object-reaching task where a seven-degree-of-freedom arm is controlled in velocity mode to reach a blue cuboid in clutter through visual observations from a monocular RGB camera. The adversarial transfer approach reduced the labeled real data requirement by 50%. Policies can be transferred to real environments with only 93 labeled and 186 unlabeled real images. The transferred visuo-motor policies are robust to novel (not seen in training) objects in clutter and even a moving target, achieving a 97.8% success rate and 1.8 cm control accuracy. Datasets and code are openly available.

Robotic pick-and-place of novel objects in clutter with multi-affordance grasping and cross-domain image matching

This article presents a robotic pick-and-place system that is capable of grasping and recognizing both known and novel objects in cluttered environments. The key new feature of the system is that it handles a wide range of object categories without needing any task-specific training data for novel objects. To achieve this, it first uses an object-agnostic grasping framework to map from visual observations to actions: inferring dense pixel-wise probability maps of the affordances for four different grasping primitive actions. It then executes the action with the highest affordance and recognizes picked objects with a cross-domain image classification framework that matches observed images to product images. Since product images are readily available for a wide range of objects (e.g., from the web), the system works out-of-the-box for novel objects without requiring any additional data collection or re-training. Exhaustive experimental results demonstrate that our multi-affordance grasping achieves high success rates for a wide variety of objects in clutter, and our recognition algorithm achieves high accuracy for both known and novel grasped objects. The approach was part of the MIT–Princeton Team system that took first place in the stowing task at the 2017 Amazon Robotics Challenge. All code, datasets, and pre-trained models are available online at http://arc.cs.princeton.edu/

A New Approach Based on Two-stream CNNs for Novel Objects Grasping in Clutter

Recently, many researches focus on learning to grasp novel objects, which is an important but still unsolved issue especially for service robots. While some approaches perform well in some cases, they need human labeling and can hardly be used in clutter with a high precision. In this paper, we apply a deep learning approach to solve the problem about grasping novel objects in clutter. We focus on two-fingered parallel-jawed grasping with RGBD camera. Firstly, we propose a ‘grasp circle’ method to find more potential grasps in each sampling point with less cost, which is parameterized by the size of the gripper. Considering the challenge of collecting large amounts of training data, we collect training data directly from cluttered scene with no manual labeling. Then we need to extract effective features from RGB and depth data, for which we propose a bimodal representation and use two-stream convolution neural networks (CNNs) to handle the processed inputs. Finally the experiment shows that compared to some existing popular methods, our method gets higher success rate of grasping for the original RGB-D cluttered scene.