A Novel Clustering-Based Algorithm for Solving Spatially Constrained Robotic Task Sequencing Problems

Wong, Cuebong; Mineo, Carmelo; Yang, Erfu; Yan, Xiu-Tian; Gu, Dongbing

doi:10.1109/tmech.2020.3037158

Cited by 10 publications

(7 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, robot kinematics and collision avoidance need to be considered in the context of automated, robot-based inspection. Hence, alternative solutions to the TSP such as forwarding view poses obtained through the presented RL framework to independent solutions for robotic task sequencing problems might be more effective [ 32 ].…”

Section: Discussionmentioning

confidence: 99%

A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks

Landgraf

Meese

Pabst

et al. 2021

Sensors

View full text Add to dashboard Cite

Manual inspection of workpieces in highly flexible production facilities with small lot sizes is costly and less reliable compared to automated inspection systems. Reinforcement Learning (RL) offers promising, intelligent solutions for robotic inspection and manufacturing tasks. This paper presents an RL-based approach to determine a high-quality set of sensor view poses for arbitrary workpieces based on their 3D computer-aided design (CAD). The framework extends available open-source libraries and provides an interface to the Robot Operating System (ROS) for deploying any supported robot and sensor. The integration into commonly used OpenAI Gym and Baselines leads to an expandable and comparable benchmark for RL algorithms. We give a comprehensive overview of related work in the field of view planning and RL. A comparison of different RL algorithms provides a proof of concept for the framework’s functionality in experimental scenarios. The obtained results exhibit a coverage ratio of up to 0.8 illustrating its potential impact and expandability. The project will be made publicly available along with this article.

show abstract

Section: Discussionmentioning

confidence: 99%

A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks

Landgraf

Meese

Pabst

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…The depth camera data origin was calibrated as robot TCP, using the hand-eye calibration procedure described in [41]. Collision avoidance wa ensured for all the robotic trajectories, to move from any actual robot pose to the next pose, implementing the effective solution proposed in [42]. A MATLAB-based simulation environment was developed through integrating the virtual CAD model of the camera with the virtual model of the robot.…”

Section: Stopping Criteriamentioning

confidence: 99%

Autonomous 3D geometry reconstruction through robot-manipulated optical sensors

Mineo

Cerniglia

Ricotta

et al. 2021

Int J Adv Manuf Technol

Self Cite

View full text Add to dashboard Cite

Many industrial sectors face increasing production demands and the need to reduce costs, without compromising the quality. The use of robotics and automation has grown significantly in recent years, but versatile robotic manipulators are still not commonly used in small factories. Beside of the investments required to enable efficient and profitable use of robot technology, the efforts needed to program robots are only economically viable in case of large lot sizes. Generating robot programs for specific manufacturing tasks still relies on programming trajectory waypoints by hand. The use of virtual simulation software and the availability of the specimen digital models can facilitate robot programming. Nevertheless, in many cases, the virtual models are not available or there are excessive differences between virtual and real setups, leading to inaccurate robot programs and time-consuming manual corrections. Previous works have demonstrated the use of robot-manipulated optical sensors to map the geometry of samples. However, the use of simple user-defined robot paths, which are not optimized for a specific part geometry, typically causes some areas of the samples to not be mapped with the required level of accuracy or to not be sampled at all by the optical sensor. This work presents an autonomous framework to enable adaptive surface mapping, without any previous knowledge of the part geometry being transferred to the system. The novelty of this work lies in enabling the capability of mapping a part surface at the required level of sampling density, whilst minimizing the number of necessary view poses. Its development has also led to an efficient method of point cloud down-sampling and merging. The article gives an overview of the related work in the field, a detailed description of the proposed framework and a proof of its functionality through both simulated and experimental evidences.

show abstract

“…The depth-camera data origin was calibrated as robot TCP, using the hand-eye calibration procedure described in [39]. Collision avoidance was ensured for all the robotic trajectories, to move from any actual robot pose to the next pose, implementing the effective solution proposed in [40]. A MATLAB-based simulation environment was developed through integrating the virtual CAD model of the camera with the virtual model of the robot.…”

Section: Stopping Criteriamentioning

confidence: 99%

Autonomous 3D geometry reconstruction through robot-manipulated optical sensors

Mineo

Cerniglia²,

Ricotta³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Many industrial sectors face increasing production demands and need to reduce costs, without compromising the quality. Whereas mass production relies on well-established protocols, small production facilities with small lot sizes struggle to update their highly changeable production at reasonable costs. The use of robotics and automation has grown significantly in recent years, but extremely versatile robotic manipulators are still not commonly used in small factories. Beside of the investments required to enable efficient and profitable use of robot technology, the efforts needed to program robots are only economically viable in case of large lot sizes. Generating robot programs for specific manufacturing tasks still relies on programming trajectory waypoints by hand. The use of virtual simulation software and the availability of the specimen digital models can facilitate robot programming. Nevertheless, in many cases, the virtual models are not available or there are unavoidable differences between virtual and real setups, leading to inaccurate robot programs and time-consuming manual corrections. This could be avoided by measuring the real-geometry and the position of the specimen, which creates the paradox of having to plan robot paths for surface mapping purposes, before the originally intended robot task can be approached. Previous works have demonstrated the use of robotically manipulated optical sensors to map the geometry of samples. However, the use of simple user-defined robot paths, which are not optimized to the part geometry, typically causes some areas of the samples to not be mapped with the required level of accuracy or to not be sampled at all by the optical sensor. This work presents an autonomous framework to enable adaptive surface mapping, without any previous knowledge of the part geometry being transferred to the system. The article gives an overview of the related work in the field, a detailed description of the proposed framework and a proof of its functionality through both simulated and experimental evidences.

show abstract

A Novel Clustering-Based Algorithm for Solving Spatially Constrained Robotic Task Sequencing Problems

Cited by 10 publications

References 27 publications

A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks

A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks

Autonomous 3D geometry reconstruction through robot-manipulated optical sensors

Autonomous 3D geometry reconstruction through robot-manipulated optical sensors

Contact Info

Product

Resources

About