Pose estimation of free-form objects is a crucial task towards flexible and reliable highly complex autonomous systems. Recently, methods based on range and RGB-D data have shown promising results with relatively high recognition rates and fast running times. On this line, this paper presents a feature-based method for 6D pose estimation of rigid objects based on the Point Pair Features voting approach. The presented solution combines a novel preprocessing step, which takes into consideration the discriminative value of surface information, with an improved matching method for Point Pair Features. In addition, an improved clustering step and a novel view-dependent re-scoring process are proposed alongside two scene consistency verification steps. The proposed method performance is evaluated against 15 state-of-the-art solutions on a set of extensive and variate publicly available datasets with real-world scenarios under clutter and occlusion. The presented results show that the proposed method outperforms all tested state-of-the-art methods for all datasets with an overall 6.6% relative improvement compared to the second best method.

This paper presents a design concept for gravity compensation of planar articulated robotic arms using a series of gear-slider mechanisms with springs. The spring-attached gear-slider mechanism has one degree-of-freedom (DOF) of motion, which can serve as a gear-spring module (GSM) to be installed onto the robot joints for leveraging the gravitational energy of the robot arm. The proposing GSM-based design is featured by its structure compactness, less assemblage effort, ease of modularization, and high performance for gravity compensation of articulated robotic manipulators. As a key part of the design, the stiffness of the spring in the GSM can be determined through either a design optimization or an analytical approximation to perfect balancing. The analyses on several 1-, 2-, and 3-DOF GSM-based robot arms illustrate that the analytical approximation to perfect balancing can reach nearly the same performance as provided through the design optimization. The power loss due to the gear contact is considered when evaluating the gravity compensation performance. A formula for spring stiffness correction is suggested for taking the power loss into account. An experimental study on a one-DOF GSM-based robot arm was performed, which shows that a power reduction rate of 86.5% is attained by the actuation motor when the GSM is installed on the robot arm.

Industrial robots are evolving rapidly in the manufacturing industry. There are two main techniques for programming the robots such as online and offline programming. However, the time spent on programming a new trajectory is a major challenge in deploying welding robots which makes these approaches less efficient. This paper presents a two-stage method employing multi-sensor interaction for the path planning of a welding robot. The proposed scheme enhances weld seam trajectory development and creates a highly adjustable intelligent guidance programming system for welding robots. A global stage approach utilizing the RGB-D camera which combines fast 2D object recognition and 3D reconstruction models is proposed to quickly identify the coarse trajectory. The processes of fast 2D object recognition and 3D reconstruction are carried out using a deep neural network model and stereo vision sensor module. This technique can be a better replacement for offline programming or hand gesture controls also known as the teaching trajectory, particularly for the welding robot application. The local positioning stage is then applied using the laser vision module to obtain more precise information of the local environment, guided by the coarse trajectory that was realized in the prior stage (global stage). The efficacy of the proposed system is analyzed by conducting numerous tests using an experimental setup. The experimental finding demonstrates that the suggested study has great potential to automate the welding robot in the manufacturing sector.

Random bin-picking is a prominent, useful, and challenging industrial robotics application. However, many industrial and real-world objects are planar and have oriented surface points that are not sufficiently compact and discriminative for those methods using geometry information, especially depth discontinuities. This study solves the above-mentioned problems by proposing a novel and robust solution for random bin-picking for planar objects in a cluttered environment. Different from other research that has mainly focused on 3D information, this study first applies an instance segmentation-based deep learning approach using 2D image data for classifying and localizing the target object while generating a mask for each instance. The presented approach, moreover, serves as a pioneering method to extract 3D point cloud data based on 2D pixel values for building the appropriate coordinate system on the planar object plane. The experimental results showed that the proposed method reached an accuracy rate of 100% for classifying two-sided objects in the unseen dataset, and 3D appropriate pose prediction was highly effective, with average translation and rotation errors less than 0.23 cm and 2.26°, respectively. Finally, the system success rate for picking up objects was over 99% at an average processing time of 0.9 s per step, fast enough for continuous robotic operation without interruption. This showed a promising higher successful pickup rate compared to previous approaches to random bin-picking problems. Successful implementation of the proposed approach for USB packs provides a solid basis for other planar objects in a cluttered environment. With remarkable precision and efficiency, this study shows significant commercialization potential.