Bin Li’s research while affiliated with Beijing Institute of Technology and other places

This paper describes an autonomous navigation and control system for capturing the maneuvering drones. A vision-based navigation method seeks and detects the intruding drone, then, the target trajectory is predicted by fusing onboard vision and inertial-measurement resources. The target’s relative position, velocity and acceleration are also obtained at the same time. Then, we present a modified proportional-derivative (PD) algorithm based on the estimated target states. In addition, the boundary constraints of the protected area are considered to avoid a collision. The proposed capture navigation and control system has demonstrated its efficiency both in simulation, flight experiments, and MBZIRC 2020, where our team won the Challenge I competition.

In aerial image object detection, how to efficiently detect different size objects in input images of different scales and obtain a unified multi-scale representation of the object is an important issue. Existing methods rarely consider the connection between multi-scale training and multi-scale inference, and do not well optimize the constraint of input object samples in the multi-scale training process, which limits the performance of multi-scale representation. In this study, an efficient object detection algorithm for aerial images is proposed to alleviate this problem. Firstly, we propose to use metric learning to obtain the scale representation boundary of each object class, reduce the support of indistinguishable objects at extreme scales in the training process, and enhance the effect of multi-scale representation. Secondly, indistinguishable small objects are merged into small object regions, and these regions are trained to recommend the detector to detect small objects on the following high-resolution scale. Thus, a reasonable association between multi-scale training and inference is established, and the efficiency of multi-scale inference is considerably improved. The proposed algorithm has been tested on three popular aerial image datasets, including VisDrone, DOTA and UAVDT. Experimental results show that it can improve the detection accuracy and reduce the number of processing pixels.

In this work, a new guidance law with a meaningful performance index is designed to satisfy terminal impact angle and impact time constraints based on optimal error dynamics, which can be used for salvo attacks or cooperative missions of multi-missile. The analytical solution of the proposed guidance law is a combination of trajectory shaping guidance law and an additional impact time error feedback term that is proportional to the difference between the desired and the true impact times. Trajectory shaping guidance law aims to achieve the desired terminal impact angle and zero miss distance, whereas the extra term aims to meet the desired impact time. The minimum and maximum feasible impact times that consider the seeker's field-of-view limit, terminal impact angle constraint, and missile's maneuvering acceleration limit are calculated to provide the feasible boundary range of the desired impact time. Numerical simulations of several engagement situations demonstrate the effectiveness of the proposed guidance law in the accuracy of terminal impact angle and impact time.

A novel systematic technique for gain-scheduled control based on fixed-structure synthesis is adopted to design the aerial vehicle autopilot. The gain-scheduled design can be transformed into the multi-model control problem with both controller architecture and gain-scheduled architecture defined a priori. Hidden coupling terms naturally arise in the linearized dynamics of the gain-scheduled controller when some of the state variables are also used as scheduling variables. Unlike traditional approaches that do not consider these terms, the proposed method takes the hidden coupling terms directly into account in the synthesis phase. Finally, numerical simulations are carried out to evaluate the effectiveness of the proposed methods.

A new robust backstepping controller with finite-time convergence for a class of uncertain nonlinear systems has been presented in this paper. The proposed controller is essentially a composite method, which consists of classical backstepping technique, extended state observer (ESO) and finite-time convergent differentiator (FTCD). More specifically, the ESO is used to estimate and compensate the mismatched as well as the matched uncertainties while the FTCD is adopted to compute the time derivatives of the virtual control laws and the reference signal. By effectively real-time estimation of ESO, the presented algorithm requires no information on the lumped uncertainties and the so-called high-gain controller is avoided completely. With the aid of Lyapunov theory, the finite-time stability of the closed-loop system is established in theory. An illustrative example of missile angle-of-attack autopilot design is given and some simulations are carried out to demonstrate the superiority of the proposed method.

Anew cooperative guidance law with impact time and impact angle constraints is presented for multiple flight vehicles to against the stationary target. The proposed guidance law has two projection components: the one that along the line-of-sight (LOS) ensures that all flight vehicles reach the target simultaneously with impact time constraints, another one that normal the LOS guarantees the convergence of the LOS angular rate with terminal impact angle constraints. Using the second Lyapunov stability method, we prove that the time-to-go of all flight vehicles reach at agreement in finite time if the connecting time of the communication topology is larger than the required convergent time. Nonlinear simulations demonstrate the effectiveness of the proposed law.