- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2018 15th Conference on Computer and Robot Vision (CRV) - Toronto, ON, Canada (2018.5.8-2018.5.10)] 2018 15th Conference on Computer and Robot Vision (CRV) - Learning a Bias Correction for Lidar-Only Motion Estimation
摘要： This paper presents a novel technique to correct for bias in a classical estimator using a learning approach. We apply a learned bias correction to a lidar-only motion estimation pipeline. Our technique trains a Gaussian process (GP) regression model using data with ground truth. The inputs to the model are high-level features derived from the geometry of the point-clouds, and the outputs are the predicted biases between poses computed by the estimator and the ground truth. The predicted biases are applied as a correction to the poses computed by the estimator. Our technique is evaluated on over 50 km of lidar data, which includes the KITTI odometry benchmark and lidar datasets collected around the University of Toronto campus. After applying the learned bias correction, we obtained significant improvements to lidar odometry in all datasets tested. We achieved around 10% reduction in errors on all datasets from an already accurate lidar odometry algorithm, at the expense of only less than 1% increase in computational cost at run-time.
关键词： Lidar Odometry,Gaussian Process,Motion Estimation,Bias Correction
更新于2025-09-23 15:23:52
-
Fractional Generalized Inverse Gaussian Process for Population Dynamics of Phase Singularities
摘要： In this paper, a generalized inverse Gaussian (GIG) process is studied in conjunction with the population dynamics of phase singularities (PSs). Special attention is paid to a stochastic analysis of PSs based on combined methods with (i) the theory of information geometry, (ii) the eigenvalue problem related to the double confluent Heun equation, (iii) classification of the statistics (sub-Poisson, Poisson, and super-Poisson), and (iv) fractional generalization to introduce a memory effect. The present theoretical method is applied to describe two-dimensional (2D) spiral wave turbulence in CO oxidation on a Pt surface and in the Aliev–Panfilov model. It is demonstrated that the fractional GIG process with the fractional index μ (μ = 0.5) can capture the profile of the PS number distribution and the scaling law of ω?1.5 in the power spectral density at large frequencies, which have been observed in real experiments and numerical simulations.
关键词： spiral wave turbulence,population dynamics,phase singularities,information geometry,generalized inverse Gaussian process,double confluent Heun equation,fractional generalization
更新于2025-09-23 15:21:21
-
Spatial Correlated Data Monitoring in Semiconductor Manufacturing Using Gaussian Process Model
摘要： In semiconductor manufacturing, various wafer tests are conducted in each stage. The analysis and monitoring of collected wafer testing data plays an important role in identifying potential problems and improving process yield. There exists three variation sources: lot-to-lot variation, wafer-to-wafer variation and site-to-site variation, which means the measurements cannot be considered independently. However, most existing control charts for monitoring wafer quality are based on the assumption that data are independently and identically distributed. To deal with the variations, we propose a mixed-effects model incorporating a Gaussian process to account for the variations. Based on the model, two control charts are implemented to detect anomalies of the measurements which can monitor the changes of the variations and the quality of products respectively. Simulation studies and results from real applications show that this model and control scheme is effective in estimating and monitoring the variation sources in the manufacturing process.
关键词： semiconductor manufacturing,mixed-effects model,statistical process control (SPC),Gaussian process
更新于2025-09-23 15:21:01
-
GP-SLAM: laser-based SLAM approach based on regionalized Gaussian process map reconstruction
摘要： Existing laser-based 2D simultaneous localization and mapping (SLAM) methods exhibit limitations with regard to either efficiency or map representation. An ideal method should estimate the map of the environment and the state of the robot quickly and accurately while providing a compact and dense map representation. In this study, we develop a new laser-based SLAM algorithm by redesigning the two core elements common to all SLAM systems, namely the state estimation and map construction. Utilizing Gaussian process (GP) regression, we propose a new type of map representation based on the regionalized GP map reconstruction algorithm. With this new map representation, both the state estimation method and the map update method can be completed with the use of concise mathematics. For small- or medium-scale scenarios, our method, consisting of only state estimation and map construction, demonstrates outstanding performance relative to traditional occupancy-grid-map-based approaches in both accuracy and especially efficiency. For large-scale scenarios, we extend our approach to a graph-based version.
关键词： Gaussian process,Laser-based,Simultaneous localization and mapping
更新于2025-09-23 15:19:57
-
Bayesian Optimization of a Free-Electron Laser
摘要： The Linac coherent light source x-ray free-electron laser is a complex scientific apparatus which changes configurations multiple times per day, necessitating fast tuning strategies to reduce setup time for successive experiments. To this end, we employ a Bayesian approach to maximizing x-ray laser pulse energy by controlling groups of quadrupole magnets. A Gaussian process model provides probabilistic predictions for the machine response with respect to control parameters, enabling a balance of exploration and exploitation in the search for the global optimum. We show that the model parameters can be learned from archived scans, and correlations between devices can be extracted from the beam transport. The result is a sample-efficient optimization routine, combining both historical data and knowledge of accelerator physics to significantly outperform existing optimizers.
关键词： Bayesian optimization,accelerator physics,free-electron laser,Gaussian process
更新于2025-09-23 15:19:57
-
Automated active fault detection in fouled dissolved oxygen sensors
摘要： Biofilm formation causes bias in dissolved oxygen (DO) sensors, which hamper their usage for automatic control and thereby balancing energy- and treatment efficiency. We analysed if a dataset that was generated with deliberate perturbations, can automatically be interpreted to detect bias caused by biofilm formation. We used a challenging set-up with realistic conditions that are required for a full-scale application. This included automated training (adapting to changing normal conditions) and automated tuning (setting an alarm threshold) to assure that the fault detection (FD)-methods are accessible to the operators. The results showed that automatic usage of FD-methods is difficult, especially in terms of automatic tuning of alarm thresholds when small training datasets only represent the normal conditions, i.e. clean sensors. Despite the challenging set-up, two FD-methods successfully improved the detection limit to 0.5 mg DO/L bias caused by biofilm formation. We showed that the studied dataset could be interpreted equally well by simpler FD-methods, as by advanced machine learning algorithms. This in turn indicates that the information contained in the actively generated data was more vital than its interpretation by advanced algorithms.
关键词： Receiver operating characteristics,Monitoring,One-class classification,Active fault detection,Gaussian process regression
更新于2025-09-16 10:30:52
-
Process Design of Laser Powder Bed Fusion of Stainless Steel Using a Gaussian Process-Based Machine Learning Model
摘要： In this work, a Gaussian process (GP)-based machine learning model is developed to predict the remelted depth of single tracks, as a function of combined laser power and laser scan speed in a laser powder bed fusion process. The GP model is trained by both simulation and experimental data from the literature. The mean absolute prediction error magni?ed by the GP model is only 0.6 lm for a powder bed with layer thickness of 30 lm, suggesting the adequacy of the GP model. Then, the process design maps of two metals, 316L and 17-4 PH stainless steels, are developed using the trained model. The normalized enthalpy criterion of identifying keyhole mode is evaluated for both stainless steels. For 316L, the result suggests that the DH (cid:2) 30 criterion hs should be related to the powder layer thickness. For 17-4 PH, the criterion should be revised to DH hs (cid:2) 25.
关键词： stainless steel,process design,Gaussian process,machine learning,laser powder bed fusion
更新于2025-09-12 10:27:22
-
A Hybrid Probabilistic Estimation Method for Photovoltaic Power Generation Forecasting
摘要： Because of stochastic nature of weather conditions, the predictability of photovoltaic (PV) power generation is poor. Compared with the point prediction, the probabilistic prediction of PV power generation can provide more information about the underlying uncertainties, which is beneficial to the stability and safety of grid dispatching and power system. Based on random forest (RF), fuzzy C-means (FCM), sparse Gaussian process (SPGP), improved grey wolf optimizer (IMGWO) algorithm, a hybrid probabilistic estimation method, in this paper, is proposed to predict the probability of PV power generation for every hour in one day. RF algorithm is firstly used to reduce multidimensional input variables. And according to the weather patterns, FCM method is adopted to divide data and get the similar samples. Finally, a hybrid forecasting method combines SPGP and IMGWO is applied to forecast the test data. With the simulation and experimental results, the validity and reliability of the proposed model (IMGWOSP) is verified. The results show that the proposed model has improved both accuracy and practicability, so the stability and safety of grid dispatching and power system can be improved.
关键词： PV power forecast,Spare Gaussian process regression,Probability prediction,Improved grey wolf optimizer algorithm
更新于2025-09-12 10:27:22
-
Energy-exergy modeling of solar radiation with most influencing input parameters
摘要： In this study, a new soft computing model Gaussian process regression (GPR) was evaluated for modeling the total solar radiation (TSR) and exergy (Ф) in Hakkari province (the region with the highest sunshine duration), Turkey. For this purpose, meteorological data include average, maximum and minimum temperature (Tave, Tmax, Tmin), relative humidity (H), sea level pressure (P), wind speed (W), and total sunbathing time (TST), wihch were used, and sensitivity analysis was applied for evaluating the results of TSR and Ф modeling. The results showed that all the input variables have significant impact on TSR and Ф modeling. Mean absolute percentage error and coefficient of determination (R2) for TSR and Ф predicted by GPR were 1.51–7.02% and 0.97–0.95, respectively. Application of five-fold cross validation method showed that GPR model is able to predict the TSR and Ф with a small size of data, but for more accuracy, it is suggested to use more than 70% of total data set for training the models. This research showed that GPR has a good ability for modeling the TSR and Ф with high accuracy, and so the engineers can use this method for the TSR and Ф prediction without using the solar radiation or exergy-to-energy ratio.
关键词： solar energy,Solar radiation exergy,Hakkari province of Turkey,Gaussian process regression (GPR),modeling
更新于2025-09-09 09:28:46
-
[IEEE 2017 International Conference on Optical Network Design and Modeling (ONDM) - Budapest (2017.5.15-2017.5.18)] 2017 International Conference on Optical Network Design and Modeling (ONDM) - A probabilistic approach for failure localization
摘要： This work considers the problem of fault localization in transparent optical networks. The aim is to localize single-link failures by utilizing statistical machine learning techniques trained on data that describe the network state upon current and past failure incidents. In particular, a Gaussian Process (GP) classi?er is trained on historical data extracted from the examined network, with the goal of modeling and predicting the failure probability of each link therein. To limit the set of suspect links for every failure incident, the proposed approach is complemented with the utilization of a Graph-Based Correlation heuristic. The proposed approach is tested on a dataset generated for an OFDM-based optical network, demonstrating that it achieves a high localization accuracy. The proposed scheme can be used by service providers for reducing the Mean-Time-To-Repair of the failure.
关键词： Graph-Based Correlation,OFDM,Gaussian Process,transparent optical networks,fault localization
更新于2025-09-04 15:30:14