arXiv subjects

Feng Shuang

Publications and source records attributed to Feng Shuang.

At least 19 recordsLinked to original sources

arXiv2024

Learning the Generalizable Manipulation Skills on Soft-body Tasks via Guided Self-attention Behavior Cloning Policy

Embodied AI represents a paradigm in AI research where artificial agents are situated within and interact with physical or virtual environments. Despite the recent progress in Embodied AI, it is still very challenging to learn the generalizable manipulation skills that can handle large deformation and topological changes on soft-body objects, such as clay, water, and soil. In this work, we proposed an effective policy, namely GP2E behavior cloning policy, which can guide the agent to learn the generalizable manipulation skills from soft-body tasks, including pouring, filling, hanging, excavating, pinching, and writing. Concretely, we build our policy from three insights:(1) Extracting intricate semantic features from point cloud data and seamlessly integrating them into the robot's end-effector frame; (2) Capturing long-distance interactions in long-horizon tasks through the incorporation of our guided self-attention module; (3) Mitigating overfitting concerns and facilitating model convergence to higher accuracy levels via the introduction of our two-stage fine-tuning strategy. Through extensive experiments, we demonstrate the effectiveness of our approach by achieving the 1st prize in the soft-body track of the ManiSkill2 Challenge at the CVPR 2023 4th Embodied AI workshop. Our findings highlight the potential of our method to improve the generalization abilities of Embodied AI models and pave the way for their practical applications in real-world scenarios.

cs.RO↗

arXiv2024

3D Hand Reconstruction via Aggregating Intra and Inter Graphs Guided by Prior Knowledge for Hand-Object Interaction Scenario

Recently, 3D hand reconstruction has gained more attention in human-computer cooperation, especially for hand-object interaction scenario. However, it still remains huge challenge due to severe hand-occlusion caused by interaction, which contain the balance of accuracy and physical plausibility, highly nonlinear mapping of model parameters and occlusion feature enhancement. To overcome these issues, we propose a 3D hand reconstruction network combining the benefits of model-based and model-free approaches to balance accuracy and physical plausibility for hand-object interaction scenario. Firstly, we present a novel MANO pose parameters regression module from 2D joints directly, which avoids the process of highly nonlinear mapping from abstract image feature and no longer depends on accurate 3D joints. Moreover, we further propose a vertex-joint mutual graph-attention model guided by MANO to jointly refine hand meshes and joints, which model the dependencies of vertex-vertex and joint-joint and capture the correlation of vertex-joint for aggregating intra-graph and inter-graph node features respectively. The experimental results demonstrate that our method achieves a competitive performance on recently benchmark datasets HO3DV2 and Dex-YCB, and outperforms all only model-base approaches and model-free approaches.

cs.CV↗

arXiv2023

OppLoD: the Opponency based Looming Detector, Model Extension of Looming Sensitivity from LGMD to LPLC2

Looming detection plays an important role in insect collision prevention systems. As a vital capability evolutionary survival, it has been extensively studied in neuroscience and is attracting increasing research interest in robotics due to its close relationship with collision detection and navigation. Visual cues such as angular size, angular velocity, and expansion have been widely studied for looming detection by means of optic flow or elementary neural computing research. However, a critical visual motion cue has been long neglected because it is so easy to be confused with expansion, that is radial-opponent-motion (ROM). Recent research on the discovery of LPLC2, a ROM-sensitive neuron in Drosophila, has revealed its ultra-selectivity because it only responds to stimuli with focal, outward movement. This characteristic of ROM-sensitivity is consistent with the demand for collision detection because it is strongly associated with danger looming that is moving towards the center of the observer. Thus, we hope to extend the well-studied neural model of the lobula giant movement detector (LGMD) with ROM-sensibility in order to enhance robustness and accuracy at the same time. In this paper, we investigate the potential to extend an image velocity-based looming detector, the lobula giant movement detector (LGMD), with ROM-sensibility. To achieve this, we propose the mathematical definition of ROM and its main property, the radial motion opponency (RMO). Then, a synaptic neuropile that analogizes the synaptic processing of LPLC2 is proposed in the form of lateral inhibition and attention. Thus, our proposed model is the first to perform both image velocity selectivity and ROM sensitivity. Systematic experiments are conducted to exhibit the huge potential of the proposed bio-inspired looming detector.

cs.CV↗

arXiv2023

DEVICE: Depth and Visual Concepts Aware Transformer for OCR-based Image Captioning

OCR-based image captioning is an important but under-explored task, aiming to generate descriptions containing visual objects and scene text. Recent studies have made encouraging progress, but they are still suffering from a lack of overall understanding of scenes and generating inaccurate captions. One possible reason is that current studies mainly focus on constructing the plane-level geometric relationship of scene text without depth information. This leads to insufficient scene text relational reasoning so that models may describe scene text inaccurately. The other possible reason is that existing methods fail to generate fine-grained descriptions of some visual objects. In addition, they may ignore essential visual objects, leading to the scene text belonging to these ignored objects not being utilized. To address the above issues, we propose a Depth and Visual Concepts Aware Transformer (DEVICE) for OCR-based image captinong. Concretely, to construct three-dimensional geometric relations, we introduce depth information and propose a depth-enhanced feature updating module to ameliorate OCR token features. To generate more precise and comprehensive captions, we introduce semantic features of detected visual concepts as auxiliary information, and propose a semantic-guided alignment module to improve the model's ability to utilize visual concepts. Our DEVICE is capable of comprehending scenes more comprehensively and boosting the accuracy of described visual entities. Sufficient experiments demonstrate the effectiveness of our proposed DEVICE, which outperforms state-of-the-art models on the TextCaps test set.

cs.CV↗

arXiv2022

Breakdown voltage and linear temperature drift in a single-molecule junction

Using first-principles calculations based on density functional theory combined with the non-equilibrium Green's function approach, the transport behaviors of a single-molecule junction formed by benzenedithiol connected to gold electrodes are investigated. The breakdown voltage for the model of benzenedithiol plus gold electrodes is 0.7 V, which is close to the experimental value. A linear response between the conductance and temperature (known as linear temperature drift) is found in the molecular device, which indicates that it could be used to maintain the stability of molecular circuits. Meanwhile, input and output with the same accuracies would be useful for designing multi-level circuits, which would be used to improve the resolution ratio in analog-to-digital converters. The present findings indicate that benzenedithiol-based single-molecule junctions would be promising functional units for molecular sensors.

cond-mat.mtrl-sci↗

arXiv2022

Solving DC Power Flow Problems Using Quantum and Hybrid algorithms

Power flow calculation plays an important role in planning, operation, and control of the power system. The quantum HHL algorithm can achieve theoretical exponential speedup over classical algorithms on DC power flow calculation. Since the qubit resources in the Noisy Intermediate-scale Quantum (NISQ) era are limited, it is important to discuss the performance considering this limitation. The coefficient matrix of the linear systems of equations in DC power flow problems cannot be represented perfectly by finite binary number strings, which leads to imperfect phase estimation. This work is carried out under the assumption of imperfect phase estimation. The performance of the HHL algorithm is systematically investigated with different accuracy and redundant qubits. In order to further reduce the required qubit resources, a hybrid quantum-classical algorithm is proposed. By comparing errors of the HHL and hybrid algorithms in the DC power flow calculation of the IEEE 5-bus test system, it is found that the hybrid algorithm can achieve comparable precision with fewer qubits than HHL by increasing the number of phase estimation modules, which may make the hybrid algorithm a feasible route in the NISQ era.

quant-ph↗

arXiv2021

A Weakly-Supervised Depth Estimation Network Using Attention Mechanism

Monocular depth estimation (MDE) is a fundamental task in many applications such as scene understanding and reconstruction. However, most of the existing methods rely on accurately labeled datasets. A weakly-supervised framework based on attention nested U-net (ANU) named as ANUW is introduced in this paper for cases with wrong labels. The ANUW is trained end-to-end to convert an input single RGB image into a depth image. It consists of a dense residual network structure, an adaptive weight channel attention (AWCA) module, a patch second non-local (PSNL) module and a soft label generation method. The dense residual network is the main body of the network to encode and decode the input. The AWCA module can adaptively adjust the channel weights to extract important features. The PSNL module implements the spatial attention mechanism through a second-order non-local method. The proposed soft label generation method uses the prior knowledge of the dataset to produce soft labels to replace false ones. The proposed ANUW is trained on a defective monocular depth dataset and the trained model is tested on three public datasets, and the results demonstrate the superiority of ANUW in comparison with the state-of-the-art MDE methods.

cs.CV↗

arXiv2021

Hybrid algorithms to solve linear systems of equations with limited qubit resources

The solution of linear systems of equations is a very frequent operation and thus important in many fields. The complexity using classical methods increases linearly with the size of equations. The HHL algorithm proposed by Harrow et al. achieves exponential acceleration compared with the best classical algorithm. However, it has a relatively high demand for qubit resources and the solution $\left| x \right\rangle $ is in a normalized form. Assuming that the eigenvalues of the coefficient matrix of the linear systems of equations can be represented perfectly by finite binary number strings, three hybrid iterative phase estimation algorithms (HIPEA) are designed based on the iterative phase estimation algorithm in this paper. The complexity is transferred to the measurement operation in an iterative way, and thus the demand of qubit resources is reduced in our hybrid algorithms. Moreover, the solution is stored in a classical register instead of a quantum register, so the exact unnormalized solution can be obtained. The required qubit resources in the three HIPEA algorithms are different. HIPEA-1 only needs one single ancillary qubit. The number of ancillary qubits in HIPEA-2 is equal to the number of nondegenerate eigenvalues of the coefficient matrix of linear systems of equations. HIPEA-3 is designed with a flexible number of ancillary qubits. The HIPEA algorithms proposed in this paper broadens the application range of quantum computation in solving linear systems of equations by avoiding the problem that quantum programs may not be used to solve linear systems of equations due to the lack of qubit resources.

quant-ph↗

arXiv2017

Self-to-self transitions in open quantum systems: the origin and solutions

The information of quantum pathways can be extracted in the framework of the Hamiltonian-encoding and Observable-decoding method. For closed quantum systems, only off-diagonal elements of the Hamiltonian in the Hilbert space is required to be encoded to obtain the desired transitions. For open quantum systems, environment-related terms will appear in the diagonal elements of the Hamiltonian in the Liouville space. Therefore, diagonal encodings have to be performed to differentiate different pathways, which will lead to self-to-self transitions and inconsistency of pathway amplitudes with Dyson expansion. In this work, a well-designed transformation is proposed to avoid the counter-intuitive transitions and the inconsistency, with or without control fields. A three-level open quantum system is employed for illustration, and numerical simulations show that the method are consistent with Dyson expansion.

quant-ph↗

arXiv2017

Doppler-free coherent-control spectroscopy with a colliding pair of shaped pulses

We demonstrate the use of the ultrafast spatial coherent-control method to resolve the fine-structure two-photon transitions of atomic rubidium. Counter-propagating ultrafast optical pulses with spectral phase and amplitude programmed with our optimized solutions successfully induced the two-photon transitions through $5S_{1/2}$-$5P_{1/2}$-$5D$ and $5S_{1/2}$-$5P_{3/2}$-$5D$ pathways, both simultaneously and at distinct spatial locations. Three different pulse-shaping solutions are introduced that combine amplitude shaping, which avoids direct intermediate resonances, and phase programming, which enables the remaining spectral components to be coherently interfered through the targeted transition pathways. Experiments were performed with a room-temperature vapor cell, and the results agree well with theoretical analysis.

physics.atom-ph↗

arXiv2015

Effective Subspace for Pure State Transfer Induced by Measurements

One main goal of quantum control is to steer a quantum system toward an expected state or dynamics. For measurement-induced quantum control, measurements serve as the only control, which is like the cases in quantum Zeno and anti-zeno effects. In this paper, this scenario will be investigated in a general $N$-level quantum system. It is proved that, when the intial and expected states are both pure, the control space could be reduced to an effective subspace spanned by these two states only. This result will greatly simplify the measurement-induced control strategy of an $N$-level quantum system.

quant-ph↗

arXiv2014

Non-Gaussian features from Excited Squeezed Vacuum State

In this work, we introduce a non-Gaussian quantum state named excited squeezed vacuum state (ESVS), which can be ustilized to describe quantum light field emitted from the multiphoton quantum process occurred in some restricted quantum systems. We investigate its nonclassical properties such as Wigner distribution in phase space, photon number distribution, the second-order autocorrelation and the quadrature fluctuations. By virtue of the Hilbert-Schmidt distance method, we quantify the non-Gaussianity of the ESVS. Due to the similar photon statistics, we examine the fidelity between the ESVS and the photon-subtraction squeezed vacuum state (PSSVS), and then find the optimal fidelity by monitoring the relevant parameters.

quant-ph↗

arXiv2012

Optimal coherent control of CARS: signal enhancement and background elimination

The ability to enhance resonant signals and eliminate the non-resonant background is analyzed for Coherent Anti-Stokes Raman Scattering (CARS). The analysis is done at a specific frequency as well as for broadband excitation using femtosecond pulse-shaping techniques. An appropriate objective functional is employed to balance resonant signal enhancement against non-resonant background suppression. Optimal enhancement of the signal and minimization of the background can be achieved by shaping the probe pulse alone while keeping the pump and Stokes pulses in transform-limited-form (TLF). In some cases analytical forms for the probe pulse can be found, and numerical simulations are carried out for other circumstances. It is found that a good approximate solution for the optimal pulse in the two-pulse CARS is a superposition of linear and arctangent type phases for the pump. The well-known probe delay method is shown to be a quasi-optimal scheme for background suppression. The results should provide a basis to improve the performance of CARS spectroscopy and microscopy.

quant-ph↗

arXiv2011

All Possible Coupling Schemes in XY Spin Chains for Perfect State Transfer

We investigate quantum state transfer in XY spin chains and propose a recursive procedure to construct the nonuniform couplings of these chains with arbitrary length to achieve perfect state transfer(PST). We show that this method is capable of finding all possible coupling schemes for PST. These schemes, without external control fields, only involve preengineered couplings but not dynamical control of them, so they can be simply realized experimentally. The analytical solutions provide all information for coupling design.

quant-ph↗

arXiv2009

Control of quantum dynamics by optimized measurements

Quantum measurements are considered for optimal control of quantum dynamics with instantaneous and continuous observations utilized to manipulate population transfer. With an optimal set of measurements, the highest yield in a two-level system can be obtained. The analytical solution is given for the problem of population transfer by measurement-assisted coherent control in a three-level system with a dynamical symmetry. The anti-Zeno effect is recovered in the controlled processes. The demonstrations in the paper show that suitable observations can be powerful tools in the manipulation of quantum dynamics.

quant-ph↗

arXiv2007

Observation-assisted optimal control of quantum dynamics

This paper explores the utility of instantaneous and continuous observations in the optimal control of quantum dynamics. Simulations of the processes are performed on several multilevel quantum systems with the goal of population transfer. Optimal control fields are shown to be capable of cooperating or fighting with observations to achieve a good yield, and the nature of the observations may be optimized to more effectively control the quantum dynamics. Quantum observations also can break dynamical symmetries to increase the controllability of a quantum system. The quantum Zeno and anti-Zeno effects induced by observations are the key operating principles in these processes. The results indicate that quantum observations can be effective tools in the control of quantum dynamics.

quant-ph↗

arXiv2006

Quantum control by von Neumann measurements

A general scheme is presented for controlling quantum systems using evolution driven by non-selective von Neumann measurements, with or without an additional tailored electromagnetic field. As an example, a 2-level quantum system controlled by non-selective quantum measurements is considered. The control goal is to find optimal system observables such that consecutive non-selective measurement of these observables transforms the system from a given initial state into a state which maximizes the expected value of a target operator (the objective). A complete analytical solution is found including explicit expressions for the optimal measured observables and for the maximal objective value given any target operator, any initial system density matrix, and any number of measurements. As an illustration, upper bounds on measurement-induced population transfer between the ground and the excited states for any number of measurements are found. The anti-Zeno effect is recovered in the limit of an infinite number of measurements. In this limit the system becomes completely controllable. The results establish the degree of control attainable by a finite number of measurements.

quant-ph↗

arXiv2006

Foundations for Cooperating with Control Noise in the Manipulation of Quantum Dynamics

This paper develops the theoretical foundations for the ability of a control field to cooperate with noise in the manipulation of quantum dynamics. The noise enters as run-to-run variations in the control amplitudes, phases and frequencies with the observation being an ensemble average over many runs as is commonly done in the laboratory. Weak field perturbation theory is developed to show that noise in the amplitude and frequency components of the control field can enhance the process of population transfer in a multilevel ladder system. The analytical results in this paper support the point that under suitable conditions an optimal field can cooperate with noise to improve the control outcome.

quant-ph↗