Current Tracker Research Articles

SiamATL: Online Update of Siamese Tracking Network via Attentional Transfer Learning.

Visual object tracking with semantic deep features has recently attracted much attention in computer vision. Especially, Siamese trackers, which aim to learn a decision making-based similarity evaluation, are widely utilized in the tracking community. However, the online updating of the Siamese fashion is still a tricky issue due to the limitation, which is a tradeoff between model adaption and degradation. To address such an issue, in this article, we propose a novel attentional transfer learning-based Siamese network (SiamATL), which fully exploits the previous knowledge to inspire the current tracker learning in the decision-making module. First, we explicitly model the template and surroundings by using an attentional online update strategy to avoid template pollution. Then, we introduce an instance-transfer discriminative correlation filter (ITDCF) to enhance the distinguishing ability of the tracker. Finally, we suggest a mutual compensation mechanism that integrates cross-correlation matching and ITDCF detection into the decision-making subnetwork to achieve online tracking. Comprehensive experiments demonstrate that our approach outperforms state-of-the-art tracking algorithms on multiple large-scale tracking datasets.

Testing of CO2 on-orbit fill/refill for the upgraded tracker thermal pump system in the Alpha Magnetic Spectrometer

The ability of resupplying fluid is vital to many aerospace applications for the sake of lifetime prolongation and performance maintenance, especially for deep-space explorations. Compared to conventional on-orbit filling technologies with the emphasis on efficiency maximization, fluid replenishment for two-phase cooling systems in space requires an accurate control of the transferred quantity constrained by its working principle. Under microgravity, weighing is impractical, and due to the dependent relationship between the saturated temperature and the saturated pressure, mass cannot be deduced from measurements of the two thermal properties, making the filled quantity determination challenging in space. For this reason, this article presents a novel quantitative on-orbit filling method for two-phase systems, through which the transferred mass is obtained from the density variation of the supercritical fluid in the supply vessel. Via heating the fluid to supercritical, a homogeneous state independent from gravity is achieved, and therefore the fluid density can be deducted from the temperature and the pressure. The filling process is driven by pressure difference and precisely regulated by solenoid valves and a flow restrictor situated in the transfer line, constituting two fundamental components. To direct the filling, an empirical model aiming to predict the transferred mass is established in terms of the pressure difference and transfer time. Based on that, a fill/refill architecture is deployed to transfer a quantitative amount of coolant to a delicate pumped CO2 two-phase thermal control system, the Upgraded Tracker Thermal Pump System in the Alpha Magnetic Spectrometer-02 on the International Space Station, which was designed and constructed to replace the current Tracker cooling system in 2019. Results of the test conducted in a thermal-vacuum chamber showed that the density-variation approach in the upgraded two-phase system achieved a filling accuracy better than 7%, and the empirical model showed consistency with the approach within 5%. Moreover, a criterion to assess when the two-phase cooling system needs a coolant refill was proposed according to the investigation of a two-phase fluid accumulator of the system under thermal regulation, in which the heat transfer is strongly affected by liquid-vapor distribution. Given the above, the proposed method contributes to the on-orbit filling for two-phase systems with more precise control over the transferred quantity. It also delivers valuable insights to the operational status evaluation of two-phase systems for future space missions.

Experimental study of different silicon sensor options for the upgrade of the CMS Outer Tracker

During the high-luminosity phase of the LHC (HL-LHC), planned to start in 2027, the accelerator is expected to deliver an instantaneous peak luminosity of up to 7.5×1034 cm−2 s−1. A total integrated luminosity of 0300 or even 0400 fb−1 is foreseen to be delivered to the general purpose detectors ATLAS and CMS over a decade, thereby increasing the discovery potential of the LHC experiments significantly. The CMS detector will undergo a major upgrade for the HL-LHC, with entirely new tracking detectors consisting of an Outer Tracker and Inner Tracker. However, the new tracking system will be exposed to a significantly higher radiation than the current tracker, requiring new radiation-hard sensors. CMS initiated an extensive irradiation and measurement campaign starting in 2009 to systematically compare the properties of different silicon materials and design choices for the Outer Tracker sensors. Several test structures and sensors were designed and implemented on 18 different combinations of wafer materials, thicknesses, and production technologies. The devices were electrically characterized before and after irradiation with neutrons, and with protons of different energies, with fluences corresponding to those expected at different radii of the CMS Outer Tracker after 0300 fb−1. The tests performed include studies with β sources, lasers, and beam scans. This paper compares the performance of different options for the HL-LHC silicon sensors with a focus on silicon bulk material and thickness.

Online Multi-expert Learning for Visual Tracking.

The correlation filters based trackers have achieved an excellent performance for object tracking in recent years. However, most existing methods use only one filter but ignore the information of the previous filters. In this paper, we propose a novel online multi-expert learning algorithm for visual tracking. In our proposed scheme, there are former trackers which retain the previous filters, and those trackers will give their predictions in each frame. The current tracker represents the filter of current frame, and both the current tracker and the former trackers constitute our expert ensemble. We use an adaptive Second-order Quantile strategy to learn the weights of each expert, which can take full advantage of all the experts. To simplify our model and remove some bad experts, we prune our models via a minimum entropy criterion. Finally, we propose a new update strategy to avoid the model corruption problem. Extensive experimental results on both OTB2013 and OTB2015 benchmarks demonstrate that our proposed tracker performs favorably against state-of-the-art methods.

Planar n-in-n quad module prototypes for the ATLAS ITk upgrade at HL-LHC

In order to meet the requirements of the High Luminosity LHC (HL-LHC), it will be necessary to replace the current tracker of the ATLAS experiment. Therefore, a new all-silicon tracking detector is being developed, the so-called Inner Tracker (ITk). The use of quad chip modules is intended in its pixel region. These modules consist of a silicon sensor that forms a unit along with four read-out chips. The current ATLAS pixel detector consists of planar n-in-n silicon pixel sensors. Similar sensors and four FE-I4 read-out chips were assembled to first prototypes of planar n-in-n quad modules. The main focus of the investigation of these modules was the region between the read-out chips, especially the central area between all four read-out chips. There are special pixel cells placed on the sensor which cover the gap between the read-out chips. This contribution focuses on the characterization of a non-irradiated device, including important sensor characteristics, charge collection determined with radioactive sources as well as hit efficiency measurements, performed in the laboratory and at testbeams. In addition, first laboratory results of an irradiated device are presented.

Robust Object Tracking With Discrete Graph-Based Multiple Experts.

Variations of target appearances due to illumination changes, heavy occlusions, and target deformations are the major factors for tracking drift. In this paper, we show that the tracking drift can be effectively corrected by exploiting the relationship between the current tracker and its historical tracker snapshots. Here, a multi-expert framework is established by the current tracker and its historical trained tracker snapshots. The proposed scheme is formulated into a unified discrete graph optimization framework, whose nodes are modeled by the hypotheses of the multiple experts. Furthermore, an exact solution of the discrete graph exists giving the object state estimation at each time step. With the unary and binary compatibility graph scores defined properly, the proposed framework corrects the tracker drift via selecting the best expert hypothesis, which implicitly analyzes the recent performance of the multi-expert by only evaluating graph scores at the current frame. Three base trackers are integrated into the proposed framework to validate its effectiveness. We first integrate the online SVM on a budget algorithm into the framework with significant improvement. Then, the regression correlation filters with hand-crafted features and deep convolutional neural network features are introduced, respectively, to further boost the tracking performance. The proposed three trackers are extensively evaluated on three data sets: TB-50, TB-100, and VOT2015. The experimental results demonstrate the excellent performance of the proposed approaches against the state-of-the-art methods.

Exploring the quality of latest sensor prototypes for the CMS Tracker Phase II Upgrade

The luminosity of the LHC will be increased by a factor of five to seven after the third long shutdown (LS3) scheduled in the mid of the next decade. The significant increase in luminosity along with the limitations of the current Tracker require a complete renewal of the CMS Outer Tracker, the Tracker Phase-2 Upgrade, during the LS3. New types of modules called PS and 2S modules are foreseen offering enhanced functionality and radiation hardness. Milestones in sensor R&D for the 2S modules as well as first characterization results are presented. AC-coupled silicon strip sensors of two vendors, produced on 6-inch as well as on 8-inch wafers, are considered which both are in n-on-p technology. Global as well as single strip parameters were measured providing insights into the quality of the sensors.

Efficient Multivariable Generalized Predictive Control for Sensorless Induction Motor Drives

This paper presents the design and the experimental validation of a new linear multivariable generalized predictive control for speed and rotor flux of induction motor. This control approach has been designed in the d-q rotating reference frame, and the indirect vector control has been employed. Load and flux observers, as well as the possibility of including a model-reference-adaptive-system speed estimator, have been considered in the implementation. The proposed controller not only provides enhanced dynamic performance but also guarantees compliance with physical voltage and current constraints. Hence, it ensures that the space vector pulsewidth modulation (SVPWM) always operates in the linear area and that the stator windings are not damaged due to overcurrent. Moreover, the controller includes a novel torque current tracker that allows obtaining an effective electromagnetic torque without a chattering phenomenon. Several simulation and experimental tests have been carried out, both in suitable and adverse conditions, even at zero speed zone, demonstrating that the proposed controller provides an efficient speed tracking and suggesting its use in industry.

Characterization of irradiated test structures for the CMS tracker upgrade

The CMS collaboration is currently conducting a campaign to identify radiation-hard materials for an upgrade of the CMS tracker. This upgrade is needed to be able to cope with the higher radiation background of the future HL-LHC; additionally the performance of the current tracker will be significantly degraded at the time of the upgrade, requiring a replacement. Several different test structures (TSs) and sensors have been designed for a 6 in. wafer layout. These wafers were produced by an industrial supplier (Hamamatsu Photonics K.K.) and differ by their bulk material (Float Zone, Magnetic Czochralski and CVD-Epi), thickness (from 50μm to 320μm) and N–P type doping. These TSs consist of different microelectronic devices including diodes, resistors or MOS structures. They enable the extraction of parameters which are not accessible in a silicon detector and allow the assessment of the quality of the sensors produced on the same wafer. The TSs have been irradiated with protons and neutrons to emulate the radiation damage caused by the particle fluence inside the future CMS tracker after 10 years of operation. This contribution will present measurements of non-irradiated and irradiated test structures at different fluences. The changes of the properties of the microelectronic devices will be discussed as well as the design of the TSs.

Electrical performance of a silicon micro-strip super-module prototype for the High-Luminosity LHC collider

A major luminosity upgrade for the Large Hadron Collider (LHC) proton–proton collider at CERN is planned for ∼2022 to allow for peak luminosities of ∼5×1034cm−2s−1. The current inner tracker of the ATLAS experiment will be completely replaced to guarantee excellent tracking capabilities in an environment characterized by much higher detector occupancies and harsher radiation backgrounds as compared with the current LHC. The new tracker will be an all-silicon based system, with pixel detectors in the innermost layers and silicon micro-strip layers at outer radii. The so-called super-module concept is an integration solution proposed for the strip barrel region of the future ATLAS internal tracker. It consists of several double-sided silicon micro-strip modules assembled into a common light-weight support structure. A first electrical prototype of a super-module has been produced, aimed to exercise the readout chain and to investigate the noise performance of such a complex system. In this paper the super-module prototype is described and first electrical results are presented.

Control and Performance Analysis for a Capacitor-coordinated Static Synchronous Compensator to Enhance Dynamic Compensation Capability

This article proposes a capacitor-coordinated static synchronous compensator, a combination of a relay-switched shunt capacitor and a voltage-sourced inverter based static synchronous compensator, for both steady-state and dynamic reactive power compensation purposes. The utilization of the shunt capacitor is to restore static synchronous compensator capacity such that the fast voltage control for successive transient disturbances is achievable. To attain tight regulation in AC- and DC-link voltages, a voltage-regulation-oriented current command generator derived from a synchronous reference frame is proposed. To facilitate the static synchronous compensator capacity restoration, the equivalent reactive current of the shunt capacitor is considered to modify the current command. A hysteresis current tracker is employed to take place of the conventional pulse width modulated voltage modulation, providing direct and fast current control. The validity of the proposed direct current controlled capacitor-coordinated static synchronous compensator has been verified by laboratory experiments on a 3-kVA digital signal processor based controlled prototype. The experimental results for the system subjected to successive load changes associated with the switching of the shunt capacitor are presented to demonstrate the effectiveness of the proposed system.

Robust Object Tracking with Online Multiple Instance Learning

In this paper, we address the problem of tracking an object in a video given its location in the first frame and no other information. Recently, a class of tracking techniques called "tracking by detection" has been shown to give promising results at real-time speeds. These methods train a discriminative classifier in an online manner to separate the object from the background. This classifier bootstraps itself by using the current tracker state to extract positive and negative examples from the current frame. Slight inaccuracies in the tracker can therefore lead to incorrectly labeled training examples, which degrade the classifier and can cause drift. In this paper, we show that using Multiple Instance Learning (MIL) instead of traditional supervised learning avoids these problems and can therefore lead to a more robust tracker with fewer parameter tweaks. We propose a novel online MIL algorithm for object tracking that achieves superior results with real-time performance. We present thorough experimental results (both qualitative and quantitative) on a number of challenging video clips.

Multi-Camera Tracking with Adaptive Resource Allocation

Sensor fusion for object tracking is attractive since the integration of multiple sensors and/or algorithms with different characteristics can improve performance. However, there exist several critical limitations to sensor fusion techniques: (1) the measurement cost increases typically as many times as the number of sensors, (2) it is not straightforward to measure the confidence of each source and give it a proper weight for state estimation, and (3) there is no principled dynamic resource allocation algorithm for better performance and efficiency. We describe a method to fuse information from multiple sensors and estimate the current tracker state by using a mixture of sequential Bayesian filters (e.g., particle filter)--one filter for each sensor, where each filter makes a different level of contribution to estimate the combined posterior in a reliable manner. In this framework, multiple sensors interact to determine an appropriate sensor for each particle dynamically; each particle is allocated to only one of the sensors for measurement and a different number of particles is assigned to each sensor. The level of the contribution of each sensor changes dynamically based on its prior information and relative measurement confidence. We apply this technique to visual tracking with multiple cameras, and demonstrate its effectiveness through tracking results in videos.

A silicon strip module for the ATLAS inner detector upgrade in the super LHC collider

The ATLAS detector is a general purpose experiment designed to fully exploit the discovery potential of the Large Hadron Collider (LHC) at a nominal luminosity of 10 34 cm −2 s −1. It is expected that after several years of successful data-taking, the LHC physics program will be extended by increasing the peak luminosity by one order of magnitude. For ATLAS, an upgrade scenario will imply the complete replacement of the Inner Detector (ID), since the current tracker will not provide the required performance due to cumulated radiation damage and a dramatic increase in the detector occupancy. In this paper, a proposal of a double-sided silicon micro-strip module for the short-strip region of the future ATLAS ID is presented. The expected thermal performance based upon detailed FEA simulations is discussed. First electrical results from a prototype version of the next generation readout front-end chips are also shown.

Upgrading the ATLAS barrel tracker for the super-LHC

It has been proposed to increase the luminosity of the large hadron collider (LHC) at CERN by an order of magnitude, with the upgraded machine dubbed super-LHC. The ATLAS experiment will require a new tracker for this high-luminosity operation due to radiation damage and event density. In order to cope with the order of magnitude increase in pile-up backgrounds at the higher luminosity, an all-silicon tracker is being designed. The new strip detector will use significantly shorter strips than the current silicon tracker in order to minimize the occupancy. As the increased luminosity will mean a corresponding increase in radiation dose, a new generation of extremely radiation-hard silicon detectors is required. An R&D program is underway to develop silicon sensors with sufficient radiation hardness. New front-end electronics and readout systems are being designed to cope with the higher data rates. The challenges facing the sensors and the cooling and mechanical support will be discussed. A possible tracker layout will be described.

ATLAS tracking at the super-LHC

Replacement of the ATLAS tracker is necessitated by the plans to extend LHC running and to greatly increase the luminosity after a long shutdown starting in the middle of the next decade. The proposed luminosity upgrade will require replacement of the tracker to achieve the higher granularity and greatly increased radiation hardness that an order of magnitude more collisions per beam crossing will imply. Not only will this render the current straw-tube-based tracker at outer radii (the TRT) inoperable, it will also require use of even more advanced radiation hard technologies than in the current semiconductor tracker (SCT). The use of an all-silicon tracker is regarded by most as the only viable solution using new advances in substrate technology and lower temperature operation as appropriate. Additionally, the upgrade will require minimal disturbance to the rest of the experiment and be obliged to use much of the existing services infrastructure. The many challenges posed by both the extreme environment and the severe constraints coming from the very tight tracker construction timescale are discussed in this paper.

Radiation hard silicon microstrip detectors for Tevatron experiments

The Silicon Microstrip Tracking detectors at the CDF and D0 experiments have now been operating for almost three years at Fermilab. These detectors were designed originally for an integrated luminosity of 2 fb −1 . As the expected luminosity for Run IIb at the Tevatron collider was initially envisioned to reach 15 fb −1 , radiation tolerances of both devices were revisited, culminating in proposals for new systems. With reduced expectations for total luminosity at ≈6 fb −1 , the full detector-replacement projects were terminated. The CDF detector is expected nevertheless to cope efficiently with the lower anticipated dose, however, the D0 experiment is planning a smaller-scale project: a Layer-0 (L0) upgrade of the silicon tracker (D0SMT). The new device will fit between the beam line and the inner layer of the current Tracker. Built of single-sided sensors, this upgrade is expected to perform well in the harsh radiation environment, and be able to withstand an integrated luminosity of 15 fb −1 . Prototypes of Run IIb sensors were irradiated using 10 MeV protons at the tandem Van de Graaff at the James R. McDonald Laboratory at Kansas State University. A fit to the 10 MeV proton data yields a damage parameter α p = 11 × 1 0 - 17 A cm . This is consistent with results from RD48 ( α p = 9.9 × 1 0 - 17 A cm ). The scaling of damage to 1 MeV neutron fluence uses a hardness factor ( κ ) derived from the non-ionizing components of the energy loss (NEIL). NEIL predicts a hardness factor of 3.87 for 10 MeV protons. We obtained an experimental value of this factor of 2.54, or 34% smaller than scaling predictions from NEIL.

Radiation hard silicon microstrip detectors for Tevatron experiments

The Silicon Microstrip Tracking detectors at the CDF and D0 experiments have now been operating for almost three years at Fermilab. These detectors were designed originally for an integrated luminosity of 2 fb −1. As the expected luminosity for Run IIb at the Tevatron collider was initially envisioned to reach 15 fb −1, radiation tolerances of both devices were revisited, culminating in proposals for new systems. With reduced expectations for total luminosity at ≈6 fb −1, the full detector-replacement projects were terminated. The CDF detector is expected nevertheless to cope efficiently with the lower anticipated dose, however, the D0 experiment is planning a smaller-scale project: a Layer-0 (L0) upgrade of the silicon tracker (D0SMT). The new device will fit between the beam line and the inner layer of the current Tracker. Built of single-sided sensors, this upgrade is expected to perform well in the harsh radiation environment, and be able to withstand an integrated luminosity of 15 fb −1. Prototypes of Run IIb sensors were irradiated using 10 MeV protons at the tandem Van de Graaff at the James R. McDonald Laboratory at Kansas State University. A fit to the 10 MeV proton data yields a damage parameter α p = 11 × 1 0 - 17 A cm . This is consistent with results from RD48 ( α p = 9.9 × 1 0 - 17 A cm ). The scaling of damage to 1 MeV neutron fluence uses a hardness factor ( κ ) derived from the non-ionizing components of the energy loss (NEIL). NEIL predicts a hardness factor of 3.87 for 10 MeV protons. We obtained an experimental value of this factor of 2.54, or 34% smaller than scaling predictions from NEIL.