Yan Ma

Dr. Yan Ma

Beihang University, Beijing, China
orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 December 2023 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
JYJisan Yang
Jisan Yang
JJJie Jiang
Jie Jiang
JLJian Li
Jian Li
YMYan Ma
Yan Ma
LTLingfeng Tian
Lingfeng Tian
GZGuangjun Zhang
Guangjun Zhang
The uncertainty of star sensors refers to the uncertainty of star observation and attitude determination. These uncertainty parameters are important performance criteria and also the basis for multisensor data fusion. In this article, the Fisher and matrix Fisher distributions in directional statistics are used to model the observed star vectors and star sensor attitude, respectively. Based on their mathematical properties and parameter estimation methods, a novel uncertainty estimation method for single-frame measurement of star sensors is proposed and derived. This method can estimate the concentration of the observed star vectors, filling a gap in previous studies. Through the maximum-likelihood estimation of attitude, this method can directly obtain the propagation of vector concentration to the attitude concentration. Under high-precision conditions, the conversions between the concentration parameters and covariances are also derived, and based on this, a multi-field-of-view (multi-FOV) data fusion method with uncertainty parameter weighting is proposed, where the uncertainty of installation matrices is also considered. The simulation results demonstrate the proposed uncertainty estimation method’s effectiveness, and the fusion method can effectively reduce the attitude measurement error of multi-FOV star sensors. Finally, the uncertainties of the three types of star sensors are evaluated and analyzed in a night-sky experiment, which further validates the proposed method’s practicality.

13 February 2025 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
YMYan Ma
Yan Ma
CGChuanshuo Gu
Chuanshuo Gu
JJJie Jiang
Jie Jiang
XWXinguo Wei
Xinguo Wei
DXDongyu Xie
Dongyu Xie
JLJian Li
Jian Li
Traditional radio-based navigation methods rely on communication with ground tracking networks to achieve deep space navigation. However, as deep space exploration continues to advance, the increasing communication distance leads to increased communication delays, which subsequently diminish the real-time capability and accuracy of the navigation. Furthermore, factors such as obstructions by celestial bodies exacerbate the inadequacy of radio navigation for the navigational demands of deep space exploration. The autonomous optical navigation technology, which primarily employs optical navigation sensors as the core navigation equipment, can obtain navigation information of the current carrier independently of ground tracking networks. It has demonstrated significant advantages in terms of autonomy, real-time capability, reliability, accuracy, and cost-effectiveness, making it an indispensable key navigation technology for deep space exploration. This article initially reviews the fundamental navigation principles applicable to different observation targets and the primary methods for determining navigation states during deep space exploration. It then systematically analyzes the characteristics of optical navigation for the four phases of deep space exploration, namely, transfer, capture, orbital, and landing phases. Finally, using typical missions as examples, this article focuses on the study of optical navigation sensors and algorithms for different phases. The review reveals that optical navigation sensors exhibit distinct characteristics across mission phases. Sensors in the transfer phase typically feature the narrowest field of view with the longest focal lengths and the largest apertures, while those used during the landing phase employ the opposite attributes. For the capture and orbital phases, sensors strike a balance between these extremes. Moreover, the spectrum range predominantly falls within the visible light band. In terms of optical navigation algorithms, various approaches are employed during different phases. Line-of-sight (LOS) navigation is most commonly used in the transfer phase. Compared with the transfer phase, the capture and orbital phases incorporate celestial surface feature navigation, and the landing phase primarily adopts terrain relative navigation (TRN). This article would serve as a valuable reference for promoting the development of autonomous optical navigation technology for deep space exploration.

26 March 2024 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
YMYan Ma
Yan Ma
JJJie Jiang
Jie Jiang
GWGangyi Wang
Gangyi Wang
JLJian Li
Jian Li
ZWZhen Wang
Zhen Wang
The inaccurate positioning of dynamic stars is the main challenge faced by dynamic star sensors. To reduce the star positioning error, researchers have focused on refining detection and positioning methods, but such methods still employ conventional exposure for star images. This study introduces a novel approach to reducing star positioning errors by using coded exposure. In this study, two key models for the encoded star strip are established: 1) the energy distribution model, based on the coded line-spread function and 2) the positioning error model, described by coded length factors. Based on the models, we demonstrate the principle of star positioning accuracy improvement by using coded exposure, and we derive the optimal code for minimizing positioning errors. The experimental results validate the correctness of the proposed models and show that using identical detection and positioning methods, compared with conventional exposure, the proposed coded exposure approach can decrease star positioning errors by more than 35% under the condition of 5°/s.

19 April 2024 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
YCYan Chen
Yan Chen
JJJie Jiang
Jie Jiang
YMYan Ma
Yan Ma
Semantic segmentation neural network classifies each pixel of the input image with semantic labels and it is widely used in varying domains, such as remote sensing, autonomous driving, and image analysis. However, neural networks exhibit a high demand for both parallel computing and extensive data processing capabilities, which greatly challenges resource-constrained embedded devices. Therefore, this article designs a field programmable gate array-based semantic segmentation neural network for lightweighting, parallelism, and bandwidth with a strategy of hardware-software co-design. The designed optimized ghost module halves the bandwidth requirement by reordering the internal structure. Group convolution, channel shuffle, and channel attention modules (CAMs) are used to further compress the network and improve accuracy. The segmentation network based on the designed optimized ghost module reduces computational complexity and parameters to 8.2 and 13.5 times, respectively. On the hardware side, an accelerator highly parallel in data input, processing and output is designed. The CAM is divided into two parts: 1) a parameter-intensive section and 2) a computation-intensive section, which are computed in parallel by the ZYNQ’s processing system and programmable logic sides, respectively. The accelerator reached 240 MHz frequency in our experiment and the segmentation of a $320\times 320$ input took 12.06 ms. It achieves a performance of 198.16 GOPS and consumes only 9.84 W of power for the entire board running at full load, which is suitable for embedded devices with requirements on real time and power consumption.

30 December 2023 in Measurement
Verified
Journal Article
Closed access
LTLingfeng Tian
Lingfeng Tian
JJJie Jiang
Jie Jiang
JYJisan Yang
Jisan Yang
YMYan Ma
Yan Ma

11 August 2020 in Optics Express
Verified
Journal Article
Closed access
JJJie Jiang
Jie Jiang
YMYan Ma
Yan Ma
GZGuangjun Zhang
Guangjun Zhang
A single field of view (FOV) celestial navigation system (SFCNS) is a navigation system that uses only one camera to observe stars. This system not only requires nonrefracted stars for attitude measurement but also refracted stars via the atmosphere for position measurement. Different from traditional star sensors, the ability of SFCNS to observe a sufficient number of nonrefracted stars and refracted stars is not only related to the camera parameters used by the system but also to the observation attitude of the system. Moreover, its measurement accuracy depends on these parameters. For the SFCNS, both a reasonable distribution of the area ratio of the refraction and nonrefraction regions in the image and a proper design of camera parameters are important. Thus, a single-FOV-double-region global parameter optimization model is established to describe the relationship between the observation attitude, camera parameters, union star detection probability, and measurement error. The smallest measurement error and the maximum union star detection probability are set as the optimization objects. Through optimization, an optimal design scheme is provided. Eventually, a simulation experiment is conducted, proving that the proposed optimal design scheme has a union detection probability of 61%, an attitude measurement error of 0.57′′, and an orbit height measurement error below 100 m.