Wu, L, Falque, R, Perez-Puchalt, V, Liu, L, Pietroni, N & Vidal-Calleja, T 2020, 'Skeleton-Based Conditionally Independent Gaussian Process Implicit Surfaces for Fusion in Sparse to Dense 3D Reconstruction', IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 1532-1539.View/Download from: Publisher's site
Liu, L, Zhang, T, Leighton, B, Zhao, L, Huang, S & Dissanayake, G 2019, 'Robust Global Structure From Motion Pipeline With Parallax on Manifold Bundle Adjustment and Initialization', IEEE Robotics and Automation Letters, vol. 4, no. 2, pp. 2164-2171.View/Download from: Publisher's site
© 2016 IEEE. In this letter, we present a novel global structure from motion (SfM) pipeline that is particularly effective in dealing with low-parallax scenes and camera motion collinear with the features that represent the environment structure. It is therefore particularly suitable in Urban SLAM, in which frequent road-facing motion poses many challenges to conventional SLAM algorithms. Our pipeline includes a recently explored bundle adjustment (BA) method that exploits a feature parameterization using Parallax angle between on-Manifold observation rays (PMBA). It is demonstrated that this BA stage has a consistently stable optimization configuration for features with any parallax and therefore low-parallax features can stay in reconstruction without pre-filtering. To allow practical usage of PMBA, we provide a compatible initialization stage in the SfM to initialize all camera poses simultaneously, exhibiting friendliness to collinear motion. This is achieved by simplifying PMBA into a hybrid graph problem of high connectivity yet small node set size, solved using a robust linear programming technique. Using simulations and a series of publicly available real datasets including "KITTI" and "Bundle Adjustment in the Large," we demonstrate the robustness of the position initialization stage in handling collinear motion and outlier matches, superior convergence performance of the BA stage in the presence of low-parallax features, and effectiveness of our pipeline to handle many sequential or out-of-order urban scenes.
Fryc, S, Liu, L & Vidal Calleja, T 2019, 'Efficient Pipeline for Mobile Brick Picking', https://ssl.linklings.net/conferences/acra/acra2019_proceedings/views/b…, Australasian Conference on Robotics and Automation, ARAA, Adelaide, pp. 1-8.
Autonomous mobile manipulation is gaining more and more attention for a range of application including disaster response, logistics, manufacturing and construction because removes work space limitation and allows object handling. A key challenge in mobile manipulation is the interaction between motion planning and perception that will deliver stable and efficient solutions. In this work, we are interested in the problem of picking a single brick shaped object from an unstructured pile using a mo- bile manipulator and a 3D camera system. We propose a robust multi-stage pipeline for a efficient, collision-free brick picking given the object pose. The key contribution of this work is a scoring function used to find the most suitable configuration considering the integrated kinematic chain of mobile base and manipulator arm. Realistic simulation results show the proposed pipeline has 100% success rate as opposed to a standard off-the-shelf solution, which has high-failure rates.
Vu, TL, Liu, L, Paul, G & Vidal Calleja, T 2019, 'Rectangular-shaped object recognition and pose estimation', ACRA 2019 Proceedings, Australian Conference on Robotics and Automation, ARAA, Adelaide, pp. 1-9.
This paper presents a novel solution for rectangular-shaped object pose estimation in the robotic bin-picking problem, using data from a single RGB-D camera collecting point cloud data from a fixed position. The key benefit of the presented framework is its ability to accurately and robustly locate an object position and orientation, which allows for high precision robotic grasping and placing of such objects in an open-loop motion execution system. Firstly, intelligent grasping surface selection is performed, then Principal Component Analysis is used for pose estimation and finally, rotation averaging is integrated to significantly
improve noise-reduction over time. Comparisons between the resulting poses and ones estimated by a traditional Iterative Closest Point
technique, have demonstrated the framework's advantages for pose estimation tasks.
Liu, L, Zhang, T, Liu, Y, Leighton, B, Zhao, L, Huang, S & Dissanayake, G 2018, 'Parallax Bundle Adjustment on Manifold with Improved Global Initialization', Springer Proceedings in Advanced Robotics (SPAR), International Workshop on the Algorithmic Foundations of Robotics, Springer, Mérida, México.
In this paper we present a novel extension to the parallax feature based
bundle adjustment (BA). We take parallax BA into a manifold form (PMBA)
along with an observation-ray based objective function. This formulation faithfully mimics the projective nature in a camera's image formation, resulting in a stable optimization configuration robust to low-parallax features. Hence it allows use of fast Dogleg optimization algorithm, instead of the usual Levenberg Marquardt. This is particularly useful in urban SLAM in which diverse outdoor environments and collinear motion modes are prevalent. Capitalizing on these properties, we propose a global initialization scheme in which PMBA is simplified into a pose-graph problem. We show that near-optimal solution can be achieved under low-noise conditions. With simulation and a series of challenging
publicly available real datasets, we demonstrate PMBA's superior convergence performance in comparison to other BA methods. We also demonstrate, with the "Bundle Adjustment in the Large" datasets, that our global initialization process successfully bootstrap the full BA in mapping many sequential or out-of-order urban scenes.
Liu, L, Wang, Y, Zhao, L & Huang, S 2017, 'Evaluation of Different SLAM Algorithms using Google Tangle Data', 2017 12th IEEE Conference on Industrial Electronics and Applications (ICIEA), IEEE Conference on Industrial Electronics and Applications, IEEE, Siem Reap, Cambodia..View/Download from: Publisher's site
In this paper, we evaluate three state-of-the-art Simultaneous Localization and Mapping (SLAM) methods using data extracted from a state-of-the-art device for indoor navigation - the Google Tango tablet. The SLAM algorithms we investigated include Preintegration Visual Inertial Navigation System (VINS), ParallaxBA and ORB-SLAM. We first describe the detailed process of obtaining synchronized IMU and image data from the Google Tango device, then we present some of the SLAM results obtained using the three different SLAM algorithms, all with the datasets collected from Tango. These SLAM results are compared with that obtained from Tango's inbuilt motion tracking system. The advantages and failure modes of the different SLAM algorithms are analysed and illustrated thereafter. The evaluation results presented in this paper are expected to provide some guidance on further development of more robust SLAM algorithms for robotic applications.
Paul, G, Liu, L & Liu, D 2016, 'A Novel Approach to Steel Rivet Detection in Poorly Illuminated Steel Structural Environments', Control, Automation, Robotics and Vision (ICARCV), 2016 14th International Conference on, International Conference on Control, Automation, Robotics and Vision, IEEE, Phuket, Thailand.View/Download from: Publisher's site
It is becoming increasingly achievable for steel
bridge structures, which are normally both inaccessible and
hazardous for humans, to be inspected and maintained by
autonomous robots. Steel bridges have been traditionally constructed
by securing plate members together with rivets. However,
rivets present a challenge for robots both in terms of cleaning and
surface traversal. This paper presents a novel approach to RGBD
image and point cloud analysis that enables rivets to be rapidly
and robustly located using low cost, non-contact sensing devices
that can be easily affixed to a robot. The approach performs
classification based on: (a) high-intensity blobs in color images,
(b) the non-linear perturbations in depth images, and (c) surface
normal clusters in 3D point clouds. The predicted rivet locations
from the three classifiers are combined using a probabilistic
occupancy mapping technique. Experiments are conducted in
several different lab and real-world steel bridge environments,
where there is no external lighting infrastructure, and the sensors
are attached to a mobile platform, i.e. a climbing inspection robot.
The location of rivets within 2m of the robot can be robustly
located within 10mm of their correct location. The state of voxels
can be predicted with above 95% accuracy, in approximately 1
second per frame.
Paul, G, Mao, S, Liu, L & Xiong, R 2015, 'Mapping Repetitive Structural Tunnel Environments for a Biologically Inspired Climbing Robot', Assistive Robots: Proceedings of the 18th International Conference on CLAWAR 2015, International Conference on Climbing and Walking Robots, World Scientific, Hangzhou, China, pp. 325-333.View/Download from: Publisher's site
This paper presents an approach to using noisy and incomplete depth-camera datasets to detect
reliable surface features for use in map construction for a caterpillar-inspired climbing robot.
The approach uses a combination of plane extraction, clustering and template matching techniques to
infer from the restricted dataset a usable map. This approach has been tested in both laboratory
and real-world steel bridge tunnel datasets generated by a climbing robot, with the results showing
that the generated maps are accurate enough for use in localisation and step trajectory planning.
Ranasinghe, R, Dantanarayana, L, Tran, A, Lie, S, Behrens, M & Liu, L 2014, 'Smart Hoist: An Assistive Robot to Aid Carers', Proceedings for the International Conference on Control Automation Robotics & Vision, International Conference on Control, Automation, Robotics and Vision, IEEE, Singapore, pp. 1285-1291.View/Download from: Publisher's site
Assistive Robotics(AR) is a rapidly expanding field, implementing advanced intelligent machines which are able to work collaboratively with a range of human users; as assistants, tools and as companions. These AR devices can assist stretched carers at residential aged care facilities to safely enhance their capacity and to improve the quality of care services. The research work presented in this paper describes the pre- liminary outcomes of a design, development and implementation of a patient lifting AR device (Smart Hoist) to reduce lower back injuries to carers while transferring patients in aged care facilities. The proposed solution, a modified conventional lifter device which consists of several sensors capable of interacting with the Smart Hoist operator and its environment, and a set of powered wheels. This solution helps carers to manoeuvre the Smart Hoist safely and intuitively. Preliminary results collected from an evaluation of the Smart Hoist conducted at the premises of IRT Woonona residential care facility confirm the improved safety, comfort and confidence for the carers.