Blinding and blurring the multi-object tracker with adversarial perturbations.

Adversarial attack reveals a potential imperfection in deep models that they are susceptible to being tricked by imperceptible perturbations added to images. Recent deep multi-object trackers combine the functionalities of detection and association, rendering attacks on either the detector or the association component an effective means of deception. Existing attacks focus on increasing the frequency of ID switching, which greatly damages tracking stability, but is not enough to make the tracker completely ineffective. To fully explore the potential of adversarial attacks, we propose Blind-Blur Attack (BBA), a novel attack method based on spatio-temporal motion information to fool multi-object trackers. Specifically, a simple but efficient perturbation generator is trained with the blind-blur loss, simultaneously making the target invisible to the tracker and letting the background be regarded as moving targets. We take TraDeS as our main research tracker, and verify our attack method on other excellent algorithms (i.e., CenterTrack, FairMOT, and ByteTrack) on MOT-Challenge benchmark datasets (i.e., MOT16, MOT17, and MOT20). BBA attack reduced the MOTA of TraDeS and ByteTrack from 69.1 and 80.3 to -238.1 and -357.0, respectively, indicating that it is an efficient method with a high degrees of transferability.

Vibration rejection of phased array telescope systems via disturbance-propagation-characteristics-based feedforward control.

Vibration rejection is one of the key techniques to stabilize the line of sight (LOS) for phased array telescope systems. Conventionally, feedback control based on image sensors is mainly used to correct the tip/tilt errors caused by disturbances and to keep the LOS stable. However, it is restricted by the sampling rate and time delay of image sensors, leading to a limited closed-loop bandwidth. Disturbances in the middle and high frequencies are hard to suppress. In this paper, disturbance-propagation-characteristics-based feedforward control is proposed to overcome these problems. A theoretical imaging model of the phased array telescope is developed to analyze the LOS disruption caused by disturbance. In addition, to improve the disturbance suppression bandwidth and correction accuracy of the system, the disturbance propagation characteristics of the phased array telescope system are analyzed. Combined with the disturbance feedforward, targeted compensation is achieved for the sub-apertures. Finally, a comparative experiment is carried out based on the self-developed Fizeau phased array telescope system to verify the superiority of the proposed method.

Demonstration of a Fizeau Directly-Imaging Sparse-Aperture Telescope with Pointing and Tracking Capabilities.

At present, the majority of sparse-aperture telescopes (SATs) are unable to observe moving targets. In this paper, we describe the construction of and present the results obtained using a Fizeau directly-imaging sparse-aperture telescope (FDISAT) that permits pointing and the tracking of moving targets. The telescope comprises three sub-apertures, each of which is equipped with a Risley prism system that permits a maximum tracking range of 5° and has independent boresight adjustment capability. On targets in various positions, experiments with pointing and tracking are conducted. The maximum root-mean-square error (RMSE) of pointing in the sub-apertures was found to be 8.22 arcsec. When considering a target moving at 0.01°/s for approximately 320 s, the maximum RMSE of tracking in the sub-apertures was found to be 4.23 arcsec. The images obtained from the focal plane detector exhibit clear interference fringes while tracking. The experimental results demonstrate that the system can effectively track moving targets, providing a method for SAT observation of moving targets.

Phase Alignment of an Array Optical Telescope System Using Balanced Detection.

Differential phase shift keying (DPSK) modulation and multi-aperture receiving are effective means for suppressing flickering, deviation, and fragmentation of the light spot by atmospheric turbulence. What is challenging in coherent beam combination of such an array receiver system is to detect and compensate for phase deviation of sub-apertures. In this paper, a method of phase alignment of an array optical telescope system using balanced detection was proposed and demonstrated. The improved Mach Zehnder Interferometer (MZI) can demodulate the digital signal and recover the phase difference at the same time. It also brings a 3 dB gain to the receiver and improves the detection sensitivity of the system. Adequate simulations with OptiSystem and MATLAB were carried out to show that the power value remains near the ideal state of 2.75 mW, and the bit error rate is less than 10-9 after phase compensation, which indicates the effectiveness and accuracy of the proposed method. Furthermore, taking the communication interruption difference of ninety degrees as an example, the system bit error rate was reduced from 1 to 10-35, and communication was established again.

Robust Risley prism control based on disturbance observer for system line-of-sight stabilization.

In this paper, a robust control based on disturbance observer is proposed to improve the tracking accuracy of the Risley prism system (RPS). Applying the flexible thin-wall ring mechanism in the RPS causes a series of tracking and pointing challenges. Disturbances such as friction, shaft deformation, and model perturbation significantly deteriorate the tracking and pointing accuracy of the RPS. Two different observer-based control methods are proposed to guarantee the tracking precision of the RPS. Moreover, the disturbance observation and compensation (DOC) performance of the proposed methods is analyzed and compared. Finally, simulation and experiment results indicate that the proposed control methods, especially the DOC-expanded state observer control mode, obtain the best performance for disturbance rejection in the RPS.

Disturbance compensation of a multiaperture imaging system based on a coupling rotating prism using an improved model compensation control.

In this paper, a cascade double-loop control (DLC) combined with modeling compensation methods is proposed to improve the tracking precision of the multiaperture imaging system (MAIS). The application of the flexible thin-wall ring mechanism in the coupling rotating prism (CRP) system causes a series of tracking and pointing challenges. Disturbances such as friction, shaft deformation, and model perturbation significantly deteriorate the tracking and pointing accuracy of the CRP. Two different modeling compensation methods that are interfaced with classical DLC are proposed to guarantee the tracking precision of the MAIS. Moreover, the disturbance observation and compensation performance of two different modeling compensation methods are analyzed and compared. Finally, simulation and experiment results indicate that the proposed control methods, especially model compensation based on speed close-loop control, obtain the best performance for disturbance rejection in the MAIS.