December 2023
·
6 Reads
What is this page?
This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.
Publications (349)
August 2023
·
50 Reads
·
2 Citations
![Figure 2. The images and poses are used as inputs for NeRF [30] to generate a 3D rendering of the captured object. The standard set of images have a lower resolution in a unit area as compared to the images taken closer to surface. The differences in results are not too noticeable from the far viewpoint, but the differences can clearly be seen from the close viewpoint.](https://www.researchgate.net/publication/369168308/figure/fig2/AS:11431281126111158@1678546973759/The-images-and-poses-are-used-as-inputs-for-NeRF-30-to-generate-a-3D-rendering-of-the_Q320.jpg)
March 2023
·
59 Reads
·
2 Citations
Robotics
High-quality maps are pertinent to performing tasks requiring precision interaction with the environment. Current challenges with creating a high-precision map come from the need for both high pose accuracy and scan accuracy, and the goal of reliable autonomous performance of the task. In this paper, we propose a multistage framework to create a high-precision map of an environment which satisfies the targeted resolution and local accuracy by an autonomous mobile robot. The proposed framework consists of three steps. Each step is intended to aid in resolving the challenges faced by conventional approaches. In order to ensure the pose estimation is performed with high accuracy, a globally accurate coarse map of the environment is created using a conventional technique such as simultaneous localization and mapping or structure from motion with bundle adjustment. The high scan accuracy is ensured by planning a path for the robot to revisit the environment while maintaining a desired distance to all occupied regions. Since the map is to be created with targeted metrics, an online path replanning and pose refinement technique is proposed to autonomously achieve the metrics without compromising the pose and scan accuracy. The proposed framework was first validated on the ability to address the current challenges associated with accuracy through parametric studies of the proposed steps. The autonomous capability of the proposed framework was been demonstrated successfully in its use for a practical mission.
November 2022
·
10 Reads
·
2 Citations
August 2022
·
17 Reads
IEEE/ASME Transactions on Mechatronics
Feature extraction and matching are the basis of many computer vision problems, such as image retrieval, object recognition, and visual odometry. In this article, we present a novel RGB-D feature with texture and geometric information (TG). It consists of a keypoint detector and a feature descriptor, which is accurate, efficient, and robust to scene variance. In the keypoint detection, we build a simplified Gaussian image pyramid to extract the texture feature. Meanwhile, the gradient of the point cloud is superimposed as the geometric feature. In the feature description, the texture information and spatial information are encoded in relative order to build a discriminative descriptor. We also construct a novel RGB-D benchmark dataset for RGB-D detector and descriptor evaluation under single variation. Comprehensive experiments are carried out to prove the superior performance of the proposed feature compared with state-of-the-art algorithms. The experimental results also demonstrate that our TG can achieve better performance especially on accuracy and the computational efficiency, making it more suitable for the real-time applications, e.g., visual odometry.
January 2022
·
18 Reads
January 2022
·
23 Reads
September 2021
·
25 Reads
·
2 Citations
September 2021
·
427 Reads
·
6 Citations
June 2021
·
64 Reads
·
13 Citations
IEEE Transactions on Robotics
This article considers simultaneous localization and mapping (SLAM) problem for robots in situations where accurate estimates for some of the robot poses, termed anchors, are available. These may be acquired through external means, for example, by either stopping the robot at some previously known locations or pausing for a sufficient period of time to measure the robot poses with an external measurement system. The main contribution is an efficient algorithm for selecting a fixed number of anchors from a set of potential poses that minimizes estimated error in the SLAM solution. Based on a graph-topological connection between the D-optimality design metric and the tree-connectivity of the pose-graph, the anchor selection problem can be formulated approximately as a submatrix selection problem for reduced weighted Laplacian matrix, leading to a cardinality-constrained submodular maximization problem. Two greedy methods are presented to solve this submodular optimization problem with a performance guarantee. These methods are complemented by Cholesky decomposition, approximate minimum degree permutation, order reuse, and rank-1 update that exploit the sparseness of the weighted Laplacian matrix. We demonstrate the efficiency and effectiveness of the proposed techniques on public-domain datasets, Gazebo simulations, and real-world experiments.
Citations (77)
... Mobile manipulators were first used for exploration in remote environments on the surface of celestial bodies [1]. In recent decades, with the proliferation of robotic systems, mobile manipulators have found use in several domains to aid human workers such as construction [2], agriculture [3], additive manufacturing [4], telehealth [5], mapping [6], [7], etc. Mobile manipulators have also been used in areas that humans can not safely access in disaster related tasks by observing the scene to create three dimensional environments [8] or measuring environmental conditions, such as radiation [9]. New companies have even been formed to push mobile manipulators to low level consumers in offices or homes [10]. ...
- Citing Article
- Full-text available
March 2023
Robotics
... Mobile manipulators were first used for exploration in remote environments on the surface of celestial bodies [1]. In recent decades, with the proliferation of robotic systems, mobile manipulators have found use in several domains to aid human workers such as construction [2], agriculture [3], additive manufacturing [4], telehealth [5], mapping [6], [7], etc. Mobile manipulators have also been used in areas that humans can not safely access in disaster related tasks by observing the scene to create three dimensional environments [8] or measuring environmental conditions, such as radiation [9]. New companies have even been formed to push mobile manipulators to low level consumers in offices or homes [10]. ...
- Citing Conference Paper
November 2022
... In addition to traditional early warning systems, new and emerging technologies have the potential to make these detection systems even better at reducing train collisions with Asian elephants. For example, machine learning algorithms can process large amounts of data from cameras and other sensors to accurately detect elephants and other wildlife in real-time (Gunasekara et al., 2021). It should be noted that sensor technologies require constant maintenance to ensure effectiveness and reliability, which can be costly and prove difficult in remote areas. ...
- Citing Conference Paper
- Full-text available
September 2021
... The motion tracking of multilink systems has found broad applications and interest in robotics [2,3,4,5,6,7], human health monitoring and therapeutics [8,9,10,11], and the like. Recently, there has been increasing interest in the motion tracking of highly dynamic multi-link systems, particularly in the field of sports [12,13,14] and vehicular safety [15,16,17,18]. Such systems are subject to significant accelerations, so additional thought is necessary to handle the tracking of high-speed motion. ...
- Citing Conference Paper
September 2021
... In addition to feature selection in VIO, submodular optimization has been used in SLAM to prune LiDAR keyframes [8], generate sparse pose graphs [5], [6], and select navigation anchor points [22]. Streaming submodular algorithms [23], [24] have been used to summarize large image datasets, but have not been used for LiDAR scan summarization or SLAM to the best of our knowledge. ...
- Citing Article
June 2021
IEEE Transactions on Robotics
... However, this assumption falters in repetitive, symmetric scenes. DSOM (Chen et al., 2021a) uses deep networks to directly regress the posterior distribution of the system's pose to guide particle sampling and proposes a trusty mechanism to adaptively adjust the particles' weights. SeqPolar (Tao et al., 2022) first extracts features from polar images projected from point clouds using a sophisticated algorithm, then retrieves the vehicle's approximate location using HMM2 (second-order hidden Markov model), and finally achieves accurate localization through point cloud registration. ...
- Citing Article
February 2021
IEEE Robotics and Automation Letters
... Ultimately, the ergodic control is designed for a very general set of mobile sensor applications, while our work is catered for the source seeking problem. Bayesian Inference and Optimization: The recent advances in Bayesian learning have inspired many source seeking studies to adopt the Bayesian methods [30], [31], [32]. These studies view the environment as a field characterized by an (unknown) density function related to measurement and use a Gaussian process or other likelihood models as a surrogate to guide the sensor movements for new measurement collections. ...
- Citing Conference Paper
October 2020
... Although it is offline, it allows users to inspect which part of their trajectory has high pose error so that later they can adjust the environment settings accordingly for better QoS, for example, adding more light and visual features at the place where the estimated pose error is high. Other works [11,12,50] have explored SLAM uncertainty quantification of pose estimates (rather than pose tracking error magnitude) using the maximum likelihood estimator covariance matrix, extracting information from the SLAM pipeline's optimization of the environment map. The work closest to ours, Ali et al. [2], demonstrated that a random forest regression model based on input sensor data characteristics can be used to estimate the average pose error over an entire trajectory. ...
- Citing Article
January 2021
IEEE Transactions on Robotics
... It is important to highlight that, instead of the aforementioned image capture method, other alternatives can be used such as a picamera connected to a raspberry [83,84], omnidirectional cameras [85] and mobile robots equipped with these or with other types of sensors [86][87][88] to detect the landmarks. If a more robust system is needed, it is even possible to use a pattern recognition algorithm to identify doors without the need for landmarks [89][90][91] and board this system on an AGV. ...
- Citing Conference Paper
- Full-text available
August 2020
... Recently, Gaussian Process Regression (GPR) has received significant interest as a technique for discovering ST data correlations. GPR provides a fundamental framework for nonlinear non-parametric Bayesian inference widely used in soil organic matter mapping [2], temperature mapping [3] and leakage detection [4]. The use of non-parametric models opens possibilities for mapping solutions to remain generic and flexible, since hyperparameters are able to be adjusted to create more accurate practical models for some specific applications. ...
- Citing Article
August 2020
Robotica
