Fei Li

Dr. Fei Li


Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

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

Representative Publications

View All Publications
1 August 2022 in Spanish Journal of Agricultural Research
Verified
Journal Article
Open access
FLFei Li
Fei Li
JLJin Lei
Jin Lei
WWWeibing Wang
Weibing Wang
SBSong Bao
Song Bao
Aim of study: To develop a novel four-claw seedling pick-up mechanism to integrate penetration and clamping into one process, realizing picking up seedlings stably and efficiently. Material and methods: A brushless DC servo motor characterized by small size, large torque, and high control precision was adopted to realize precise control for speed and clamping force through control algorithms. An optimization model was derived according to the requirements of picking up seedlings. The parameter optimization of the seedling pick-up mechanism is a complex multi-parameter and nonlinear optimization problem. The hybrid PSO-SA algorithm was used to solve the model, and the optimal parameters of the mechanism were obtained. Main results: The dynamic simulation was conducted, and the results showed that the optimized mechanism could meet the trajectory, acceleration, and clamping force requirement for picking up seedlings. The performance tests showed that the success ratio in picking up seedlings reached 84.46%, indicating the feasibility of the four-claw seedling pick-up mechanism for automatic transplanting machines. Research highlights: The four-claw seedling pick-up mechanism can be used in the automatic transplanting machine. Additionally, the hybrid PSO-SA algorithm can achieve promising results in solving the optimization problem of the seedling pick-up mechanism.

25 August 2022 in Agriculture
Verified
Journal Article
Open access
PDF file
XLXiaojun Li
Xiaojun Li
WWWeibing Wang
Weibing Wang
GLGanghui Liu
Ganghui Liu
RLRunze Li
Runze Li
FLFei Li
Fei Li
In greenhouse nurseries, one of the important tasks of the automatic transplanter is replanting missing or bad seedling holes with healthy seedlings. This requires the transplanter to spend significant time moving between the supply trays and target trays during replanting. The diversity and complexity of the transplanting routes affect transplanter efficiency. Path planning method can find a better path for the manipulator and improve the efficiency of transplantation. The A* algorithm (A*), which is one of the optimal path search algorithms, is often used in practical applications of path planning. In this paper, the heuristic function of the A* is optimized by the ant colony algorithm (ACA), and an improved A* algorithm (Imp-A*) is obtained. Simulation tests and transplanting trials of Imp-A*, A*, ACA, Dijkstra (DA), and common sequence method (CSM) were carried out using 32-, 50-, 72-, and 128-hole plug trays. The results show that Imp-A* inherits the advantages of A* and ACA in terms of path planning length and computation time. Compared to A*, ACA, DA, and CSM, the transplanting time for Imp-A* was reduced by 2.4%, 12.84%, 11.63%, and 14.27%, respectively. In just six trays of transplanting tasks, Imp-A* saves 60.91 s compared to CSM, with an average time saving of 10.15 s per tray. The combination optimization algorithm has similar application prospects in agriculture.

27 September 2025 in Fusion Engineering and Design
Verified
Journal Article
Closed access
FLFei Li
Fei Li
WCWusheng Chou
Wusheng Chou

2 January 2024 in Fusion Engineering and Design
Verified
Journal Article
Closed access
FLFei Li
Fei Li
PHPan Hongtao
Pan Hongtao
CBChristian Bachmann
Christian Bachmann
GJGünter Janeschitz
Günter Janeschitz
YCYong Cheng
Yong Cheng
WCWusheng Chou
Wusheng Chou

27 December 2025 in Sensors
Verified
Journal Article
Open access
PDF file
FLFei Li
Fei Li
WCWusheng Chou
Wusheng Chou
Accurate distance perception and collision reasoning are crucial for robotic manipulation in the confined interior of tokamak vacuum vessels. Traditional mesh- or voxel-based methods suffer from discretization artifacts, discontinuities, and heavy memory requirements, making them unsuitable for continuous geometric reasoning and optimization-based planning. This paper presents an Occupancy-Aware Neural Distance Perception (ONDP) framework that serves as a compact and differentiable geometric sensor for manipulator obstacle avoidance in reactor-like environments. To address the inadequacy of conventional sampling methods in such constrained environments, we introduce a Physically-Stratified Sampling strategy. This approach moves beyond heuristic adaptation to explicitly dictate data distribution based on specific engineering constraints. By injecting weighted quotas into critical safety buffers and enforcing symmetric boundary constraints, we ensure robust gradient learning in high-risk regions. A lightweight neural network is trained directly in physical units (millimeters) using a mean absolute error loss, ensuring strict adherence to engineering tolerances. The resulting model achieves approximately 2–3 mm near-surface accuracy and supports high-frequency distance and normal queries for real-time perception, monitoring, and motion planning. Experiments on a tokamak vessel model demonstrate that ONDP provides continuous, sub-centimeter geometric fidelity. Crucially, benchmark results confirm that the proposed method achieves a query frequency exceeding 15 kHz for large-scale batches, representing a 5911× speed-up over mesh-based queries. This breakthrough performance enables its seamless integration with trajectory optimization and model-predictive control frameworks for confined-space robotic manipulation.