arXiv subjects

Kejun Li

Publications and source records attributed to Kejun Li.

28 records · Page 2Linked to original sources

arXiv2022

Safety-Aware Preference-Based Learning for Safety-Critical Control

Bringing dynamic robots into the wild requires a tenuous balance between performance and safety. Yet controllers designed to provide robust safety guarantees often result in conservative behavior, and tuning these controllers to find the ideal trade-off between performance and safety typically requires domain expertise or a carefully constructed reward function. This work presents a design paradigm for systematically achieving behaviors that balance performance and robust safety by integrating safety-aware Preference-Based Learning (PBL) with Control Barrier Functions (CBFs). Fusing these concepts -- safety-aware learning and safety-critical control -- gives a robust means to achieve safe behaviors on complex robotic systems in practice. We demonstrate the capability of this design paradigm to achieve safe and performant perception-based autonomous operation of a quadrupedal robot both in simulation and experimentally on hardware.

cs.RO↗

arXiv2022

Nuclear spin polarization and control in a van der Waals material

Van der Waals layered materials are a focus of materials research as they support strong quantum effects and can easily form heterostructures. Electron spins in van der Waals materials played crucial roles in many recent breakthroughs, including topological insulators, two-dimensional (2D) magnets, and spin liquids. However, nuclear spins in van der Waals materials remain an unexplored quantum resource. Here we report the first demonstration of optical polarization and coherent control of nuclear spins in a van der Waals material at room temperature. We use negatively-charged boron vacancy ($V_B^-$) spin defects in hexagonal boron nitride to polarize nearby nitrogen nuclear spins. Remarkably, we observe the Rabi frequency of nuclear spins at the excited-state level anti-crossing of $V_B^-$ defects to be 350 times larger than that of an isolated nucleus, and demonstrate fast coherent control of nuclear spins. We also detect strong electron-mediated nuclear-nuclear spin coupling that is 5 orders of magnitude larger than the direct nuclear spin dipolar coupling, enabling multi-qubit operations. Nitrogen nuclear spins in a triangle lattice will be suitable for large-scale quantum simulation. Our work opens a new frontier with nuclear spins in van der Waals materials for quantum information science and technology.

quant-ph↗

arXiv2022

Natural Multicontact Walking for Robotic Assistive Devices via Musculoskeletal Models and Hybrid Zero Dynamics

Generating stable walking gaits that yield natural locomotion when executed on robotic-assistive devices is a challenging task that often requires hand-tuning by domain experts. This paper presents an alternative methodology, where we propose the addition of musculoskeletal models directly into the gait generation process to intuitively shape the resulting behavior. In particular, we construct a multi-domain hybrid system model that combines the system dynamics with muscle models to represent natural multicontact walking. Provably stable walking gaits can then be generated for this model via the hybrid zero dynamics (HZD) method. We experimentally apply our integrated framework towards achieving multicontact locomotion on a dual-actuated transfemoral prosthesis, AMPRO3, for two subjects. The results demonstrate that enforcing muscle model constraints produces gaits that yield natural locomotion (as analyzed via comparison to motion capture data and electromyography). Moreover, gaits generated with our framework were strongly preferred by the non-disabled prosthetic users as compared to gaits generated with the nominal HZD method, even with the use of systematic tuning methods. We conclude that the novel approach of combining robotic walking methods (specifically HZD) with muscle models successfully generates anthropomorphic robotic-assisted locomotion.

cs.RO↗

arXiv2021

$C_2C_N$ as a 2 eV Single-Photon Emitter Candidate in Hexagonal Boron Nitride

The generation of single-photon emitters in hexagonal boron nitride around 2 eV emission is experimentally well-recognized; however the atomic nature of these emitters is unknown. In this paper, we use first-principles calculations to demonstrate that $C_2C_N$ is a possible source of 2 eV single-photon emitter. We showcase the calculations of a complete set of static and dynamical properties related to defects, including exciton-defect couplings and electron-phonon interactions. In particular, we show it is critical to consider nonradiative processes when comparing with experimental lifetime for known 2 eV emitters. We find that $C_2C_N$ has several key physical properties matching the ones of experimentally observed single-photon emitters. These include the zero-phonon line (2.13eV), Huang-Rhys factor (1.35), photoluminescence lifetime (2.19 ns), phonon-sideband energy (180meV), and photoluminescence spectrum. The identification of defect candidates for 2 eV emission paves the way for controllable single-photon emission generation.

cond-mat.mtrl-sci↗

arXiv2021

ROIAL: Region of Interest Active Learning for Characterizing Exoskeleton Gait Preference Landscapes

Characterizing what types of exoskeleton gaits are comfortable for users, and understanding the science of walking more generally, require recovering a user's utility landscape. Learning these landscapes is challenging, as walking trajectories are defined by numerous gait parameters, data collection from human trials is expensive, and user safety and comfort must be ensured. This work proposes the Region of Interest Active Learning (ROIAL) framework, which actively learns each user's underlying utility function over a region of interest that ensures safety and comfort. ROIAL learns from ordinal and preference feedback, which are more reliable feedback mechanisms than absolute numerical scores. The algorithm's performance is evaluated both in simulation and experimentally for three non-disabled subjects walking inside of a lower-body exoskeleton. ROIAL learns Bayesian posteriors that predict each exoskeleton user's utility landscape across four exoskeleton gait parameters. The algorithm discovers both commonalities and discrepancies across users' gait preferences and identifies the gait parameters that most influenced user feedback. These results demonstrate the feasibility of recovering gait utility landscapes from limited human trials.

cs.RO↗

arXiv2021

Intersystem Crossing and Exciton-Defect Coupling of Spin Defects in Hexagonal Boron Nitride

Despite the recognition of two-dimensional (2D) systems as emerging and scalable host materials of single photon emitters or spin qubits, uncontrolled and undetermined chemical nature of these quantum defects has been a roadblock to further development. Leveraging the design of extrinsic defects can circumvent these persistent issues and provide an ultimate solution. Here we established a complete theoretical framework to accurately and systematically design quantum defects in wide-bandgap 2D systems. With this approach, essential static and dynamical properties are equally considered for spin qubit discovery. In particular, many-body interactions such as defect-exciton couplings are vital for describing excited state properties of defects in ultrathin 2D systems. Meanwhile, nonradiative processes such as phonon-assisted decay and intersystem crossing rates require careful evaluation, which compete together with radiative processes. From a thorough screening of defects based on first-principles calculations, we identify promising single photon emitters such as SiVV and spin qubits such as TiVV and MoVV in hexagonal boron nitride. This work provided a complete first-principles theoretical framework for defect design in 2D materials.

cond-mat.mtrl-sci↗

arXiv2019

Automatic quality assessment for 2D fetal sonographic standard plane based on multi-task learning

The quality control of fetal sonographic (FS) images is essential for the correct biometric measurements and fetal anomaly diagnosis. However, quality control requires professional sonographers to perform and is often labor-intensive. To solve this problem, we propose an automatic image quality assessment scheme based on multi-task learning to assist in FS image quality control. An essential criterion for FS image quality control is that all the essential anatomical structures in the section should appear full and remarkable with a clear boundary. Therefore, our scheme aims to identify those essential anatomical structures to judge whether an FS image is the standard image, which is achieved by three convolutional neural networks. The Feature Extraction Network aims to extract deep level features of FS images. Based on the extracted features, the Class Prediction Network determines whether the structure meets the standard and Region Proposal Network identifies its position. The scheme has been applied to three types of fetal sections, which are the head, abdominal, and heart. The experimental results show that our method can make a quality assessment of an FS image within less a second. Also, our method achieves competitive performance in both the detection and classification compared with state-of-the-art methods.

cs.CV↗

arXiv2019

Characteristics of solar wind rotation

Over 54 years of hourly mean value of solar wind velocity from 27 Nov. 1963 to 31 Dec. 2017 are used to investigate characteristics of the rotation period of solar wind through auto-correlation analysis. Solar wind of high velocity is found to rotate faster than low-velocity wind, while its rotation rate increases with velocity increasing, but in contrast for solar wind of low velocity, its rotation rate decreases with velocity increasing. Our analysis shows that solar wind of a higher velocity statistically possesses a faster rotation rate for the entire solar wind. The yearly rotation rate of solar wind velocity does not follow the Schwable cycle, but it is significantly negatively correlated to yearly sunspot number when it leads by 3 years. Physical explanations are proposed to these findings.

astro-ph.SR↗

arXiv2010

Relationship between group sunspot number and Wolf sunspot number

Continuous wavelet transform and cross-wavelet transform have been used to investigate the phase periodicity and synchrony of the monthly mean Wolf ($R_{z}$) and group ($R_{g}$) sunspot numbers during the period of June 1795 to December 1995. The Schwabe cycle is the only one common period in Rg and Rz, but it is not well-defined in case of cycles 5-7 of Rg and in case of cycles 5 and 6 of $R_{z}$. In fact, the Schwabe period is slightly different in $R_{g}$ and $R_{z}$ before cycle 12, but from cycle 12 onwards it is almost the same for the two time series. Asynchrony of the two time series is more obviously seen in cycles 5 and 6 than in the following cycles, and usually more obviously seen around the maximum time of a cycle than during the rest of the cycle. $R_{g}$ is found to fit $R_{z}$ better in both amplitudes and peak epoch during the minimum time time of a solar cycle than during the maximum time of the cycle, which should be caused by their different definition, and around the maximum time of a cycle, $R_{g}$ is usually less than $R_{z}$. Asynchrony of $R_{g}$ and $R_{z}$ should somewhat agree with different sunspot cycle characteristics exhibited by themselves.

astro-ph.SR↗

arXiv2010

The Phase Shifts of the Paired Wings of Butterfly Diagrams

Sunspot groups observed by Royal Greenwich Observatory/US Air Force/NOAA from May 1874 to November 2008 and the Carte Synoptique solar filaments from March 1919 to December 1989 are used to investigate the relative phase shift of the paired wings of butterfly diagrams of sunspot and filament activities. Latitudinal migration of sunspot groups (or filaments) does asynchronously occur in the northern and southern hemispheres, and there is a relative phase shift between the paired wings of their butterfly diagrams in a cycle, making the paired wings spatially asymmetrical on the solar equator. It is inferred that hemispherical solar activity strength should evolve in a similar way within the paired wings of a butterfly diagram in a cycle, making the paired wings just and only keep the phase relationship between the northern and southern hemispherical solar activity strengths, but a relative phase shift between the paired wings of a butterfly diagram should bring about an almost same relative phase shift of hemispheric solar activity strength.

astro-ph.SR↗