Search

Your search keyword '"Gokul, B."' showing total 126 results
Start Over Author "Gokul, B." Publication Year Range Last 10 years
126 results on '"Gokul, B."'

1. Enhancing Visual Place Recognition via Fast and Slow Adaptive Biasing in Event Cameras

Author

Nair, Gokul B., Milford, Michael, and Fischer, Tobias

Subjects
Computer Science - Robotics, Computer Science - Computer Vision and Pattern Recognition
Abstract

Event cameras are increasingly popular in robotics due to beneficial features such as low latency, energy efficiency, and high dynamic range. Nevertheless, their downstream task performance is greatly influenced by the optimization of bias parameters. These parameters, for instance, regulate the necessary change in light intensity to trigger an event, which in turn depends on factors such as the environment lighting and camera motion. This paper introduces feedback control algorithms that automatically tune the bias parameters through two interacting methods: 1) An immediate, on-the-fly \textit{fast} adaptation of the refractory period, which sets the minimum interval between consecutive events, and 2) if the event rate exceeds the specified bounds even after changing the refractory period repeatedly, the controller adapts the pixel bandwidth and event thresholds, which stabilizes after a short period of noise events across all pixels (\textit{slow} adaptation). Our evaluation focuses on the visual place recognition task, where incoming query images are compared to a given reference database. We conducted comprehensive evaluations of our algorithms' adaptive feedback control in real-time. To do so, we collected the QCR-Fast-and-Slow dataset that contains DAVIS346 event camera streams from 366 repeated traversals of a Scout Mini robot navigating through a 100 meter long indoor lab setting (totaling over 35km distance traveled) in varying brightness conditions with ground truth location information. Our proposed feedback controllers result in superior performance when compared to the standard bias settings and prior feedback control methods. Our findings also detail the impact of bias adjustments on task performance and feature ablation studies on the fast and slow adaptation mechanisms., Comment: 8 pages, 9 figures, paper accepted to the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2024)

Published
2024

4. Optimizing The Cut And Count Method In Phenomenological Studies

Author

Coleppa, Baradhwaj, Krishna, Gokul B., Sarkar, Agnivo, and Shil, Sujay

Subjects
High Energy Physics - Phenomenology
Abstract

We introduce an optimization technique to discriminate signal and background in any phenomenological study based on the cut and count-based method. The core ideas behind this algorithm is the introduction of a ranking scheme that can quantitatively assess the relative importance of various observables involved in a new physics process, and a more methodical way of choosing what cuts to impose. The technique is an iterative process that works with the help of the MadAnalysis5 interface. Working in the context of a simple BSM scenario with a heavy Higgs boson that decays to four leptons via ZZ, we demonstrate how automating the cut and count process in this specific way results in an enhanced discovery potential compared with the more traditional way of imposing cuts.

Published
2023

7. Charged Higgs Prospects In Extended Gauge Models

Author

Coleppa, Baradhwaj, Krishna, Gokul B., and Sarkar, Agnivo

Subjects
High Energy Physics - Phenomenology
Abstract

In this paper, we explore the collider phenomenology of the charged Higgs boson in the context of a generic Beyond Standard Model scenario with extended gauge and scalar sectors. In such scenarios, the charged Higgs boson can decay via the $W^{'}Z/ WZ^{'}$ channels. We formulate a search strategy for the $H^{\pm}$ in the channel $\sigma(g b \rightarrow H^{\pm}t)\mathcal{BR}(H^{\pm} \rightarrow W' Z)$ considering the interesting cascade decay chain $H^{\pm} \rightarrow W^{'} Z \rightarrow W^{\pm} Z Z$. We find that the charged Higgs can be discovered in final states with multiple hard leptons and/ or b-quarks which future LHC experiments with sufficiently large luminosity ($\mathcal{L} = 1000 fb^{-1}$ and above) can probe.

Published
2021

8. BirdSLAM: Monocular Multibody SLAM in Bird's-Eye View

Author

Daga, Swapnil, Nair, Gokul B., Ramesh, Anirudha, Sajnani, Rahul, Ansari, Junaid Ahmed, and Krishna, K. Madhava

Subjects
Computer Science - Robotics, Computer Science - Artificial Intelligence, Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning
Abstract

In this paper, we present BirdSLAM, a novel simultaneous localization and mapping (SLAM) system for the challenging scenario of autonomous driving platforms equipped with only a monocular camera. BirdSLAM tackles challenges faced by other monocular SLAM systems (such as scale ambiguity in monocular reconstruction, dynamic object localization, and uncertainty in feature representation) by using an orthographic (bird's-eye) view as the configuration space in which localization and mapping are performed. By assuming only the height of the ego-camera above the ground, BirdSLAM leverages single-view metrology cues to accurately localize the ego-vehicle and all other traffic participants in bird's-eye view. We demonstrate that our system outperforms prior work that uses strictly greater information, and highlight the relevance of each design decision via an ablation analysis., Comment: Accepted in VISIGRAPP (VISAPP) 2021

Published
2020

9. Banana Ripeness Identification and Classification Using Hybrid Models with RESNET-50, VGG-16 and Machine Learning Techniques

Author

Mathew, Milan Bins, Surya Manjunathan, G., Gokul, B., Mohana Ganesh, K., Sajidha, S. A., Nisha, V. M., Angrisani, Leopoldo, Series Editor, Arteaga, Marco, Series Editor, Panigrahi, Bijaya Ketan, Series Editor, Chakraborty, Samarjit, Series Editor, Chen, Jiming, Series Editor, Chen, Shanben, Series Editor, Chen, Tan Kay, Series Editor, Dillmann, Rüdiger, Series Editor, Duan, Haibin, Series Editor, Ferrari, Gianluigi, Series Editor, Ferre, Manuel, Series Editor, Hirche, Sandra, Series Editor, Jabbari, Faryar, Series Editor, Jia, Limin, Series Editor, Kacprzyk, Janusz, Series Editor, Khamis, Alaa, Series Editor, Kroeger, Torsten, Series Editor, Li, Yong, Series Editor, Liang, Qilian, Series Editor, Martín, Ferran, Series Editor, Ming, Tan Cher, Series Editor, Minker, Wolfgang, Series Editor, Misra, Pradeep, Series Editor, Möller, Sebastian, Series Editor, Mukhopadhyay, Subhas, Series Editor, Ning, Cun-Zheng, Series Editor, Nishida, Toyoaki, Series Editor, Oneto, Luca, Series Editor, Pascucci, Federica, Series Editor, Qin, Yong, Series Editor, Seng, Gan Woon, Series Editor, Speidel, Joachim, Series Editor, Veiga, Germano, Series Editor, Wu, Haitao, Series Editor, Zamboni, Walter, Series Editor, Zhang, Junjie James, Series Editor, Sisodia, Dilip Singh, editor, Garg, Lalit, editor, Pachori, Ram Bilas, editor, and Tanveer, M., editor

Published
2023
Full Text
View/download PDF

10. Multi-object Monocular SLAM for Dynamic Environments

Author

Nair, Gokul B., Daga, Swapnil, Sajnani, Rahul, Ramesh, Anirudha, Ansari, Junaid Ahmed, Jatavallabhula, Krishna Murthy, and Krishna, K. Madhava

Subjects
Computer Science - Robotics, Computer Science - Computer Vision and Pattern Recognition
Abstract

In this paper, we tackle the problem of multibody SLAM from a monocular camera. The term multibody, implies that we track the motion of the camera, as well as that of other dynamic participants in the scene. The quintessential challenge in dynamic scenes is unobservability: it is not possible to unambiguously triangulate a moving object from a moving monocular camera. Existing approaches solve restricted variants of the problem, but the solutions suffer relative scale ambiguity (i.e., a family of infinitely many solutions exist for each pair of motions in the scene). We solve this rather intractable problem by leveraging single-view metrology, advances in deep learning, and category-level shape estimation. We propose a multi pose-graph optimization formulation, to resolve the relative and absolute scale factor ambiguities involved. This optimization helps us reduce the average error in trajectories of multiple bodies over real-world datasets, such as KITTI. To the best of our knowledge, our method is the first practical monocular multi-body SLAM system to perform dynamic multi-object and ego localization in a unified framework in metric scale., Comment: Accepted to IEEE Intelligent Vehicles Symposium 2020 (IV2020)

Published
2020

11. Charged Higgs prospects in extended gauge models

Author

Baradhwaj Coleppa, Gokul B. Krishna, and Agnivo Sarkar

Subjects
Astrophysics, QB460-466, Nuclear and particle physics. Atomic energy. Radioactivity, QC770-798
Abstract

Abstract In this paper, we explore the collider phenomenology of charged Higgs bosons in the context of a generic Beyond Standard Model scenario with extended gauge and scalar sectors. In such scenarios, the charged Higgs boson can decay via the $$W^{'}Z/ WZ^{'}$$ W ′ Z / W Z ′ channels. We formulate a search strategy for the $$H^{\pm }$$ H ± in the channel $$\sigma (g b \rightarrow H^{\pm }t){\mathcal{B}\mathcal{R}}(H^{\pm } \rightarrow W' Z)$$ σ ( g b → H ± t ) B R ( H ± → W ′ Z ) considering the interesting cascade decay chain $$H^{\pm } \rightarrow W^{'} Z \rightarrow W^{\pm } Z Z$$ H ± → W ′ Z → W ± Z Z . We find that the charged Higgs can be discovered in final states with multiple hard leptons and/or b-quarks which future LHC experiments with sufficiently large luminosity ( $${\mathcal {L}} = 1000~\text {fb}^{-1}$$ L = 1000 fb - 1 and above) can probe. We then study the phenomenological implications of our findings in the context of a generic toy model with a $$H^{\pm }$$ H ± as well as in a more realistic electroweak symmetry breaking scenario.

Published
2023
Full Text
View/download PDF

18. DEM Simulation of Finite Size Particle Impact on a Loose Granular Bed

Author

Gokul, B., Patil, Dhiraj V., Cavas-Martínez, Francisco, Series Editor, Chaari, Fakher, Series Editor, Gherardini, Francesco, Series Editor, Haddar, Mohamed, Series Editor, Ivanov, Vitalii, Series Editor, Kwon, Young W., Series Editor, Trojanowska, Justyna, Series Editor, di Mare, Francesca, Series Editor, Prabu, T., editor, Viswanathan, P., editor, Agrawal, Amit, editor, and Banerjee, Jyotirmay, editor

Published
2021
Full Text
View/download PDF

23. Enhanced H2 sensing performance of perforated 3D NiO nanostructures and their comparative study with green synthesized NiO nanoparticles.

Author

Thilakavathi, G., Raja, Annamalai, Blessy, W. Hannah, Gokul, B., Parthasarathy, M., Sangeetha, J. E., Cholan, S., Kang, Misook, and El-marghany, Adel

Abstract

In this work, perforated 3D NiO nanostructures have been synthesized by polyol method and their H2 gas sensing performance was studied. To understand H2 sensing performance of 3D perforated NiO nanostructures, it is compared with green synthesized NiO (GNiO) and Polyol-assisted hydrothermal synthesized (HSNiO) NiO nanostructures. Green synthesis method resulted in nanoparticle morphology, polyol-assisted hydrothermal method as 3D microsphere, and polyol method gave perforated 3D microspheres. XRD shows polycrystalline nature of NiO with grain size 15 – 22 nm. BET surface analysis clearly depicts that surface area of perforated 3D nanostructures (SNiO) is twice and thrice than HSNiO and GNiO nanostructures. XPS analysis confirmed the presence of Ni3+/Ni2+, high adsorption of O−, and oxygen vacancies due to high surface area. All three nanostructures were tested towards different H2 gas concentrations resulted in 69%, 58% and 31% for SNiO, HSNiO and GNiO nanostructures for 100 ppm at 240 °C, respectively. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

24. An experimental study on compressive strength of M30 concrete using 15% of novel E-waste as a partial replacement of coarse aggregate on comparing with conventional concrete.

Author

Gokul, B. and Rajkumar, J. Samuel Simron

Subjects
*COMPRESSIVE strength, *STANDARD deviations, *ANALYSIS of variance, *CONFIDENCE intervals, *CONCRETE
Abstract

The primary purpose of this study is to evaluate the similarities and differences between conventional concrete and the concrete material that is composed of 15 percent Novel E-waste and is of M30 grade. The Components and Procedures: Using a combination of superplasticizer, cement (53 grade PPC), fine aggregate (sieve size 4.75mm), coarse aggregate (sieve size 20mm), and water, our research was conducted using novel electronic waste with a size of 20 millimetres. This was done in accordance with the findings of the literature review. It was decided to take into account two different groupings. The experimental group, which consisted of novel E-waste concrete, was compared to the control group, which consisted of traditional concrete. In all, there are 36 samples, with 18 samples belonging to the first group and 18 samples belonging to the second group. In order to carry out the computation, a G-power of 0.8 is utilised, along with alpha and beta quality of 0.05 and 0.2, and a confidence interval of 95 percent is utilised. The independent sample T-test value is less than 0.05, which is equivalent to p=0.004. Results: One-way Using the SPSS programme, an analysis of variance was performed to determine the compressive strength of conventional concrete and Novel E-waste Concrete. Concrete that is conventional has a compressive strength value of 31.12 N/mm2 on average. The concrete made from novel electronic trash has a compressive strength of 37.86 N/mm2. A significant value of 0.000 (p<0.05) and standard deviations of 0.65 and 1.26 are shown, respectively, in this study. The conclusion is that the compressive strength of the novel E-waste concrete with the superplasticizer was somewhat lower than that of the conventional concrete. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

25. Partial replacement of coarse aggregate in M30 grade of concrete with 20% of novel e-waste.

Author

Gokul, B. and Rajkumar, J Samuel Simron

Subjects
*CONCRETE mixing, *ONE-way analysis of variance, *ELECTRONIC waste, *COMPRESSIVE strength, *STANDARD deviations
Abstract

The purpose of the research is to demonstrate that the amount of novel electronic trash that is substituted in concrete mix at a rate of twenty percent should be higher than the conventional concrete material used for the M30 grade of concrete. The Components and Procedures: As part of the Modified Concrete Mix, the following components are included: water, superplasticizer, cement (53 grade PPC), fine aggregate (sieve size 4.75mm), coarse aggregate (sieve size 20mm), and novel electronic waste with a 20mm particle size. Results: According to the findings that were described, the compressive strength of conventional concrete and Novel E-waste Concrete was revealed by using SPSS to conduct a one-way analysis of variance. Concrete that is conventional has a compressive strength value of 31.12 N/mm2 on average. There is a compressive strength of 40.34 N/mm2 for concrete made from novel electronic trash. It is possible to carry out the computation by making use of a G-power of 0.8, alpha and beta quality values of 0.05 and 0.2, and a confidence interval equivalent to 95 percent. The outcome of the T-test, together with the significant difference value having a value of 0.000 (p<0.05), with the standard deviations being 0.65 and 1.35 respectively. When compared to traditional concrete, the compressive strength of the novel E-waste concrete that contains the superplasticizer is significantly higher. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

26. Modification of Novel E-waste in M30 grade of concrete.

Author

Gokul, B., Rajkumar, J. S. S., and Thiruchelvam, V.

Subjects
*ELECTRONIC waste, *COMPRESSIVE strength, *ONE-way analysis of variance, *STANDARD deviations, *CONFIDENCE intervals
Abstract

The scope of the research is to compare conventional concrete and the 10% Novel E-waste replaced concrete material of M30 grade concrete. The composition of various items present in concrete mix are segregated with Novel E-waste 20mm, superplasticizer, cement (53 grade PPC), fine aggregate (sieve size 4.75mm), coarse aggregate (sieve size 20mm), and water. In this research the main set of events are grouped in two categories. Group-1 (control group/conventional concrete) and group-2 (experimental group/Novel E-waste concrete). In this project first group has 18 samples and the Second group has 18 samples, Total samples of N=36 The analysis was performed with a G-power of 0.8, alpha and beta values set at 0.05 and 0.2 respectively, and a 95% confidence interval. The independent sample T-Test yielded a value of (p>0.05), specifically (p=0.256). The One-way ANOVA conducted using SPSS confirmed differences in the mechanical properties between conventional concrete and Novel E-waste Concrete. The average compressive strength for conventional concrete was found to be (31.12 N/mm2), while for Novel E-waste concrete it was (34.15 N/mm2). A significant difference was noted with a significance value of (p<0.05), precisely (p=0.000), and standard deviations of 0.65 and 0.79, respectively. It was determined that Novel E-waste concrete, when mixed with a superplasticizer, exhibited a lower compressive strength percentage compared to conventional concrete. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

29. Software Defect Density Prediction Using Deep Learning

Author

Varsini, V. Raaga, Lakshmi, S. Abitha, Devarshi, Gokul, B., Varsini, V. Raaga, Lakshmi, S. Abitha, Devarshi, and Gokul, B.

Abstract

Software defect prediction is the use of various approaches and procedures to discover and anticipate future flaws in a software project before they become costly and disruptive problems. Organizations may use data analysis, machine learning, and historical defect data to make educated decisions about resource allocation, software testing methodologies, and, ultimately, software product quality. Software defect prediction is the technique of identifying software modules that are likely to have flaws. The suggested system is intended for software defect prediction, combining PCA with a software business analysis model and employing Rank SVM as its primary predictive modeling approach. It includes data collecting from historical defects and relevant data sources, preprocessing to clean and convert the data, feature selection to identify critical indicators, and training a PCA using a software business analysis model utilizing Rank SVM. This methodology, once installed, forecasts the risk of faults in software modules based on their characteristics. The system's output influences resource allocation and testing procedures, hence improving software quality and development efficiency. Continuous monitoring and adjustments assure continuing correctness, making this an efficient option for proactive defect control in software development.

Published
2024

43. Development of a Web Framework for E-Voting System

Author

Roopa K, Gokul B S, and Arakalgud, Kaushik S

Subjects
Django, E-voting, Paillier algorithm, MySQL, Web framework
Abstract

—In modern age, there are huge requirements for web applications like ticket booking for shows, reservation for travelling, ordering food, booking a cab, online payment transactions and so on. This paper discusses a web application designed for e-voting system using open source tools like Django web framework and MySQL database. Python language is used for programming. Voting confidentiality is provided through Paillier homomorphic algorithm. The web application is designed to provide voter, administrative and electoral support. In the implemented system, the electoral head (EH) controls the voting process and is the trusted entity. The Paillier algorithm along with its web interface was simulated and tested successfully.&nbsp

Published
2022
Full Text
View/download PDF

44. Development of a fuel injector relative performance tester and cleaner.

Author

Kumar, S. Ranjith, Santhosh, S., Kumar, M. Logesh, Aswin, V., and Gokul, B.

Subjects
INJECTORS, CENTRAL processing units, FUEL systems
Abstract

Fuel injector is considered to be the heart of a vehicles fuel circulation system. They are very much important to the overall functionality and performance of any vehicle. The main principle behind the working of a fuel injector in a vehicle is spraying or dispersion of fuel into the engine. In todays technology each and every modernized vehicle consists of an ECM which is considered to be the central processing unit for the engine and it is the vehicles engine computer. This computerized control supervises the overall frequency of dispersion of fuel into the engine with specified time intervals. It also defines the timeline of the engine by calculating the efficiency and it stimulates the overall performance of the system. Certain symptoms indicate that the fuel injector is failing and losing its efficiency. Dispersion or spraying discharge of the fuel minimizes over the time. Clogging is one of the defects that happens in the injector. This clogging will in turn affect the engine performance directly or indirectly. Scale formation is also a defect in the inner bore of the nozzle of the fuel injector which occurs due to idleness. To overcome the above concerns a fuel injector relative performance tester and cleaner was developed. By utilizing this device a fuel injectors performance can be analyzed and the nozzle passage can be cleaned. This will in turn improve the performance of the engine. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

46. A qualitative study of fourteen African countries' nursing workforce and labour market.

Author

Bryant, A, Reynolds, NR, Hart, L, Johnson, PG, Kalula, A, Gokul, B, Davidson, PM, Bryant, A, Reynolds, NR, Hart, L, Johnson, PG, Kalula, A, Gokul, B, and Davidson, PM

Abstract

Aim

The aim of this study was to describe factors affecting nursing education and labour markets in countries in East, Central, and Southern Africa, and critical areas for investment.

Background

An understanding about the relationship between the supply of nurses (determined by types of educational programmes, and the quantity and quality of nurse graduates), and workforce demand is critical to health policy development.

Methods

Six focus groups and 14 key informant interviews with nursing leaders and experts were conducted. Participants included government chief nursing officers, registrars of regulatory bodies, association leaders and heads of nursing education. The data were transcribed, coded and analysed using inductive techniques.

Findings

Participants discussed challenges and strengths of nursing education, school and regulatory infrastructure, financing mechanisms for the nursing workforce, the state of nursing jobs and scope of nursing practice.

Conclusion

Strengthened regulations and leadership are needed to improve investment in nursing, the quality of nursing education, and working conditions and to promote the achievement of better health outcomes.

Implications for nursing policy

Clarifying scope of practice for nurses in the health sector and creating competency-based requirements is important. Governments should establish positions that align with updated competencies and provide fair and safe working conditions. The current and ongoing investment case for nursing requires improved data systems and a commitment to use labour market data for decision-making.

Published
2021

49. Exchange bias and magnetocrystalline anisotropy of non-stoichiometric CoxFe3−xO4 nanoparticles.

Author

Gokul, B., Matheswaran, P., Pandian, M., Arun Paul, C., Ravikumar, K., and Abd El-Rehim, A. F.

Subjects
MAGNETIC anisotropy, MAGNETIC structure, NANOPARTICLES, MAGNETIC properties, SHOWROOMS, MAGNETITE, PHOTOELECTRONS
Abstract

We report the influence of Co concentration on the exchange bias and magnetocrystalline anisotropy of CoxFe3−xO4 (x = 0.0, 0.4, 0.7, 1.0) nanoparticles. Co ions were substituted in magnetite (Fe3O4) nanoparticles to obtain non-stoichiometric CoxFe3−xO4 nanoparticles. The concentration of Co (mol) was varied from 0.0 to 1.0. Structure and magnetic anisotropic properties of non-stoichiometric CoxFe3−xO4 nanoparticles were studied. The prepared CoxFe3−xO4 nanoparticle was quasi-spherical in shape with particle sizes ranging from 15 to 20 nm. Raman and X-ray photoelectron spectroscopic (XPS) studies suggest the transformation of inverse to normal spinel structure concerning Co concentration. Blocking temperature increases from 155 to 302 K with an increase in concentration (mol) from 0.0 to 1.0 respectively. CoxFe3−xO4 (x = 0.7) resulted in an exchange bias field and large magnetocrystalline anisotropy of about 70 Oe and 5.38 × 104 J/m3. Due to large magnetocrystalline anisotropy, CoxFe3−xO4 (x = 0.7) shows a low exchange bias field. CoxFe3−xO4 (x = 1.0) show room temperature hysteresis with coercivity of about 14.5 kOe at 5 K and MS of 45 emu/g. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

50. Multi-object Monocular SLAM for Dynamic Environments

Author

K. Madhava Krishna, Krishna Murthy Jatavallabhula, Junaid Ahmed Ansari, Anirudha Ramesh, Swapnil Daga, Rahul Sajnani, and Gokul B. Nair

Subjects
FOS: Computer and information sciences, 0209 industrial biotechnology, Computer science, media_common.quotation_subject, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition, ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION, 02 engineering and technology, Simultaneous localization and mapping, Vehicle dynamics, Computer Science - Robotics, 020901 industrial engineering & automation, 0202 electrical engineering, electronic engineering, information engineering, Computer vision, Absolute scale, media_common, Monocular, business.industry, Deep learning, Ambiguity, Object (computer science), Computer Science::Computer Vision and Pattern Recognition, Trajectory, 020201 artificial intelligence & image processing, Artificial intelligence, business, Robotics (cs.RO)
Abstract

In this paper, we tackle the problem of multibody SLAM from a monocular camera. The term multibody, implies that we track the motion of the camera, as well as that of other dynamic participants in the scene. The quintessential challenge in dynamic scenes is unobservability: it is not possible to unambiguously triangulate a moving object from a moving monocular camera. Existing approaches solve restricted variants of the problem, but the solutions suffer relative scale ambiguity (i.e., a family of infinitely many solutions exist for each pair of motions in the scene). We solve this rather intractable problem by leveraging single-view metrology, advances in deep learning, and category-level shape estimation. We propose a multi pose-graph optimization formulation, to resolve the relative and absolute scale factor ambiguities involved. This optimization helps us reduce the average error in trajectories of multiple bodies over real-world datasets, such as KITTI. To the best of our knowledge, our method is the first practical monocular multi-body SLAM system to perform dynamic multi-object and ego localization in a unified framework in metric scale., Comment: Accepted to IEEE Intelligent Vehicles Symposium 2020 (IV2020)

Published
2020
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results