Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216758
Yongliang Qiao, Daobilige Su, He Kong, S. Sukkarieh, S. Lomax, C. Clark
Accurate segmentation of cattle is a prerequisite for feature extraction and estimation. Convolutional neural networks (CNN) based approaches that train models on the largescale labeled datasets have achieved high levels of segmentation performance. However, pixel-wise manual labeling of a cattle image is challenging and time consuming due to the irregularity of the cattle contour. In this regard, data augmentation for deep learning based cattle segmentation is required. Our proposed data augmentation approach uses random image cropping and patching to expand the number of training images and their corresponding labels, then, a state-of-the-art deep neural net is trained to segment cattle images. Here we apply these techniques to images of cattle in a feedlot environment. Our data augmentation-based approach segmented cattle from a complex background with 99.5% mean Accuracy (mAcc) and 97.3% mean Intersection of Unions (mIoU), improving current techniques including a combination of random flipping, rotation and color jitter.
{"title":"Data Augmentation for Deep Learning based Cattle Segmentation in Precision Livestock Farming","authors":"Yongliang Qiao, Daobilige Su, He Kong, S. Sukkarieh, S. Lomax, C. Clark","doi":"10.1109/CASE48305.2020.9216758","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216758","url":null,"abstract":"Accurate segmentation of cattle is a prerequisite for feature extraction and estimation. Convolutional neural networks (CNN) based approaches that train models on the largescale labeled datasets have achieved high levels of segmentation performance. However, pixel-wise manual labeling of a cattle image is challenging and time consuming due to the irregularity of the cattle contour. In this regard, data augmentation for deep learning based cattle segmentation is required. Our proposed data augmentation approach uses random image cropping and patching to expand the number of training images and their corresponding labels, then, a state-of-the-art deep neural net is trained to segment cattle images. Here we apply these techniques to images of cattle in a feedlot environment. Our data augmentation-based approach segmented cattle from a complex background with 99.5% mean Accuracy (mAcc) and 97.3% mean Intersection of Unions (mIoU), improving current techniques including a combination of random flipping, rotation and color jitter.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131260787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216787
Yu Zhang, Miguel Martínez-García
This paper proposes to apply a machine hearing framework for industrial fault diagnosis, which is inspired by humans’ “listening and diagnostic” capability in identifying machinery faults. The proposed method combines simplified human auditory functionalities with machine learning, aiming to model in a more biologically plausible way. It includes primarily using cochleagram to extract useful time-frequency information in sound signals -representing the cochlea filtering properties in human hearing. Then, a recurrent neural network with long short-term memory layers is constructed to learn and classify the cochleagrams for fault diagnosis – this is to incorporate memory elements in temporal information processing. The proposed method is validated with an experimental study on bearing fault diagnosis using acoustic measurements, while the developed machine hearing scheme could be beneficial to many industrial fault diagnosis applications, e.g., for aeronautical, automotive, marine, railway and manufacturing industry.
{"title":"Machine Hearing for Industrial Fault Diagnosis","authors":"Yu Zhang, Miguel Martínez-García","doi":"10.1109/CASE48305.2020.9216787","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216787","url":null,"abstract":"This paper proposes to apply a machine hearing framework for industrial fault diagnosis, which is inspired by humans’ “listening and diagnostic” capability in identifying machinery faults. The proposed method combines simplified human auditory functionalities with machine learning, aiming to model in a more biologically plausible way. It includes primarily using cochleagram to extract useful time-frequency information in sound signals -representing the cochlea filtering properties in human hearing. Then, a recurrent neural network with long short-term memory layers is constructed to learn and classify the cochleagrams for fault diagnosis – this is to incorporate memory elements in temporal information processing. The proposed method is validated with an experimental study on bearing fault diagnosis using acoustic measurements, while the developed machine hearing scheme could be beneficial to many industrial fault diagnosis applications, e.g., for aeronautical, automotive, marine, railway and manufacturing industry.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"89 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132910258","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216732
Daqiang Guo, Shiquan Ling, Hao Li, Di Ao, Tongda Zhang, Yiming Rong, G. Huang
The booming customized and personalized demands call for new production paradigms that complies with that change. The ubiquitous connection, digitization and sharing in the context of Industry 4.0 present an opportunity for next-generation production paradigm-personalized production, to meet the booming personalized demands with individual needs and preferences. Personalized production refers to a customer-centric production paradigm, where individual needs and preferences are transformed into personalized products and services at an affordable cost, by maximizing the benefit of connection and sharing throughout the product life-cycle. This paper reviews and identifies the evolution of production paradigms. A framework for personalized production based on digital twin, blockchain and additive manufacturing in the context of Industry 4.0 is proposed. Besides, the impact of the implementation of personalized production is discussed from the aspects of customer-centric business model, social and environmental effects and challenges of data ownership. This paper provides helpful guidance and reference for personalized production paradigm.
{"title":"A framework for personalized production based on digital twin, blockchain and additive manufacturing in the context of Industry 4.0","authors":"Daqiang Guo, Shiquan Ling, Hao Li, Di Ao, Tongda Zhang, Yiming Rong, G. Huang","doi":"10.1109/CASE48305.2020.9216732","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216732","url":null,"abstract":"The booming customized and personalized demands call for new production paradigms that complies with that change. The ubiquitous connection, digitization and sharing in the context of Industry 4.0 present an opportunity for next-generation production paradigm-personalized production, to meet the booming personalized demands with individual needs and preferences. Personalized production refers to a customer-centric production paradigm, where individual needs and preferences are transformed into personalized products and services at an affordable cost, by maximizing the benefit of connection and sharing throughout the product life-cycle. This paper reviews and identifies the evolution of production paradigms. A framework for personalized production based on digital twin, blockchain and additive manufacturing in the context of Industry 4.0 is proposed. Besides, the impact of the implementation of personalized production is discussed from the aspects of customer-centric business model, social and environmental effects and challenges of data ownership. This paper provides helpful guidance and reference for personalized production paradigm.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"313 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115867444","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216879
Yixiang Liu, Qing Bi, Xizhe Zang, Yibin Li
Human walking gait is much more natural looking and energy efficient compared with biped robots. This paper presents a bio-inspired approach to realizing more human-like biped robotic walking. For this purpose, a biped robot actuated coordinately by pneumatic artificial muscles and springs is developed, capable of exploiting passive compliance of the mechanical system in locomotion. And a control scheme for human-like walking is designed based on finite state machine. Experiments show that the biped robot can walk stably on the treadmill even with some small obstacles. The realized gait has the important features of human walking, including heel strike and toe off, stretched knees, and variation in the height of the body’s center of mass, which demonstrates the effectiveness of the proposed approach.
{"title":"Human-like Walking of a Biped Robot Actuated by Pneumatic Artificial Muscles and Springs*","authors":"Yixiang Liu, Qing Bi, Xizhe Zang, Yibin Li","doi":"10.1109/CASE48305.2020.9216879","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216879","url":null,"abstract":"Human walking gait is much more natural looking and energy efficient compared with biped robots. This paper presents a bio-inspired approach to realizing more human-like biped robotic walking. For this purpose, a biped robot actuated coordinately by pneumatic artificial muscles and springs is developed, capable of exploiting passive compliance of the mechanical system in locomotion. And a control scheme for human-like walking is designed based on finite state machine. Experiments show that the biped robot can walk stably on the treadmill even with some small obstacles. The realized gait has the important features of human walking, including heel strike and toe off, stretched knees, and variation in the height of the body’s center of mass, which demonstrates the effectiveness of the proposed approach.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121435011","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9217043
M. P. Fanti, A. M. Mangini, M. Roccotelli, B. Silvestri
The hospital sector is implementing several services and procedures in order to improve the quality and assistances through new technologies. In this context, the drugs distribution is a very important activity to provide an efficient service to all departments that need supply. The spread of new viruses, such as COVID19, or other dangers, which requires the decrease of interactions between people even within the hospital sector, can also be limited using a fleet of autonomous robot vehicles. Drugs cross delivery in a hospital is an activity that can be performed through the use of these new vehicles. In this paper an innovative optimization approach of drugs cross distribution within a hospital is proposed, in order to reduce both number and length of trips and number of autonomous robot vehicles in the fleet, without significantly reducing the level of the provided service. The idea is based on the collaborative logistics concept in which a limited number of autonomous robot vehicles are used for time-scheduled delivery activities through a combination of two departments to be served for each delivery. This strategy is formalized by an Integer Linear Programming Problem to optimize the delivery tasks. Moreover, a case study simulation is presented to show the application of the methodology in a hospital.
{"title":"Hospital Drugs Distribution with Autonomous Robot Vehicles","authors":"M. P. Fanti, A. M. Mangini, M. Roccotelli, B. Silvestri","doi":"10.1109/CASE48305.2020.9217043","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9217043","url":null,"abstract":"The hospital sector is implementing several services and procedures in order to improve the quality and assistances through new technologies. In this context, the drugs distribution is a very important activity to provide an efficient service to all departments that need supply. The spread of new viruses, such as COVID19, or other dangers, which requires the decrease of interactions between people even within the hospital sector, can also be limited using a fleet of autonomous robot vehicles. Drugs cross delivery in a hospital is an activity that can be performed through the use of these new vehicles. In this paper an innovative optimization approach of drugs cross distribution within a hospital is proposed, in order to reduce both number and length of trips and number of autonomous robot vehicles in the fleet, without significantly reducing the level of the provided service. The idea is based on the collaborative logistics concept in which a limited number of autonomous robot vehicles are used for time-scheduled delivery activities through a combination of two departments to be served for each delivery. This strategy is formalized by an Integer Linear Programming Problem to optimize the delivery tasks. Moreover, a case study simulation is presented to show the application of the methodology in a hospital.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115530254","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216880
Giovanni Bianco, S. Bracco, F. Delfino, Lorenzo Gambelli, M. Robba, M. Rossi
A BEMS (Building Energy Management System) for demand response is proposed for a smart building equipped with renewables and a Heating, Ventilation and Air Conditioning System (HVAC) fed by a geothermal heat pump. The developed BEMS is based on an optimization model able to manage the HVAC plant, fan coils and rooms’ temperature to minimize costs, guarantee the desired comfort inside rooms, and track demand response signals from a DSO (Distribution System Operator). Results are provided for a real test-case represented by the Smart Energy Building (SEB) located at Savona Campus (University of Genoa, Italy).
{"title":"A Building Energy Management System for demand response in smart grids","authors":"Giovanni Bianco, S. Bracco, F. Delfino, Lorenzo Gambelli, M. Robba, M. Rossi","doi":"10.1109/CASE48305.2020.9216880","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216880","url":null,"abstract":"A BEMS (Building Energy Management System) for demand response is proposed for a smart building equipped with renewables and a Heating, Ventilation and Air Conditioning System (HVAC) fed by a geothermal heat pump. The developed BEMS is based on an optimization model able to manage the HVAC plant, fan coils and rooms’ temperature to minimize costs, guarantee the desired comfort inside rooms, and track demand response signals from a DSO (Distribution System Operator). Results are provided for a real test-case represented by the Smart Energy Building (SEB) located at Savona Campus (University of Genoa, Italy).","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124106587","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216899
Chenghao Li, Xiaoteng Ma, Li Xia, Qianchuan Zhao, Jun Yang
Traffic congestion is a severe issue of a developing world. Recently, many researchers are attempting to utilize deep reinforcement learning algorithms to bring intelligence to traffic lights. To the best of our knowledge, most prior researchers only consider the average criterion of all vehicles while training. However, fairness is another important metric but ignored. In this paper, we study the fairness control of traffic light and propose a deep reinforcement learning algorithm to optimize the fairness of all drivers’ waiting time. The objective is to minimize the maximal waiting time of drivers during a light time loop, which also partly reflects the optimization of the average waiting time. We conduct experiments for a 4-lane crossroad in SUMO. Simulation results show that our algorithm can efficiently optimize the fairness criterion. Meanwhile the average criterion is further improved. We wish to shed light on complementing the entire framework of reinforcement learning with our research on fairness control.
{"title":"Fairness Control of Traffic Light via Deep Reinforcement Learning","authors":"Chenghao Li, Xiaoteng Ma, Li Xia, Qianchuan Zhao, Jun Yang","doi":"10.1109/CASE48305.2020.9216899","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216899","url":null,"abstract":"Traffic congestion is a severe issue of a developing world. Recently, many researchers are attempting to utilize deep reinforcement learning algorithms to bring intelligence to traffic lights. To the best of our knowledge, most prior researchers only consider the average criterion of all vehicles while training. However, fairness is another important metric but ignored. In this paper, we study the fairness control of traffic light and propose a deep reinforcement learning algorithm to optimize the fairness of all drivers’ waiting time. The objective is to minimize the maximal waiting time of drivers during a light time loop, which also partly reflects the optimization of the average waiting time. We conduct experiments for a 4-lane crossroad in SUMO. Simulation results show that our algorithm can efficiently optimize the fairness criterion. Meanwhile the average criterion is further improved. We wish to shed light on complementing the entire framework of reinforcement learning with our research on fairness control.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121560879","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216740
Han Yu Li, Michael Danielczuk, A. Balakrishna, V. Satish, Ken Goldberg
The ability of robots to grasp novel objects has industry applications in e-commerce order fulfillment and home service. Data-driven grasping policies have achieved success in learning general strategies for grasping arbitrary objects. However, these approaches can fail to grasp objects which have complex geometry or are significantly outside of the training distribution. We present a Thompson sampling algorithm that learns to grasp a given object with unknown geometry using online experience. The algorithm leverages learned priors from the Dexterity Network robot grasp planner to guide grasp exploration and provide probabilistic estimates of grasp success for each stable pose of the novel object. We find that seeding the policy with the Dex-Net prior allows it to more efficiently find robust grasps on these objects. Experiments suggest that the best learned policy attains an average total reward 64.5% higher than a greedy baseline and achieves within 5.7% of an oracle baseline when evaluated over 300, 000 training runs across a set of 3000 object poses.
{"title":"Accelerating Grasp Exploration by Leveraging Learned Priors","authors":"Han Yu Li, Michael Danielczuk, A. Balakrishna, V. Satish, Ken Goldberg","doi":"10.1109/CASE48305.2020.9216740","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216740","url":null,"abstract":"The ability of robots to grasp novel objects has industry applications in e-commerce order fulfillment and home service. Data-driven grasping policies have achieved success in learning general strategies for grasping arbitrary objects. However, these approaches can fail to grasp objects which have complex geometry or are significantly outside of the training distribution. We present a Thompson sampling algorithm that learns to grasp a given object with unknown geometry using online experience. The algorithm leverages learned priors from the Dexterity Network robot grasp planner to guide grasp exploration and provide probabilistic estimates of grasp success for each stable pose of the novel object. We find that seeding the policy with the Dex-Net prior allows it to more efficiently find robust grasps on these objects. Experiments suggest that the best learned policy attains an average total reward 64.5% higher than a greedy baseline and achieves within 5.7% of an oracle baseline when evaluated over 300, 000 training runs across a set of 3000 object poses.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123907388","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9216765
Yingli Zhao, Zhengxi Hu, Lei Zhou, Meng Liu, Jingtai Liu
Group detection of crowds is an important and challenging problem in applications of the crowds analysis. Especially for service robots, accurate group detection is the premise to ensure the safe interaction between humans and robots. In this paper, we propose an online group detection method based on Structured Prediction for middle density crowds. First of all, we extend the features of pairwise trajectories with velocity and orientation to obtain more valid information. Then, a fully-connected social network is maintained to improve time efficiency significantly. Finally, we adopt the adaptive-sampling BCFW algorithm to learn the mapping from trajectories to groups. Comparing with current state-of-the-art methods, our experiments demonstrate the group detection capacity on precision and time efficiency.
{"title":"Data-driven Online Group Detection Based on Structured Prediction","authors":"Yingli Zhao, Zhengxi Hu, Lei Zhou, Meng Liu, Jingtai Liu","doi":"10.1109/CASE48305.2020.9216765","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9216765","url":null,"abstract":"Group detection of crowds is an important and challenging problem in applications of the crowds analysis. Especially for service robots, accurate group detection is the premise to ensure the safe interaction between humans and robots. In this paper, we propose an online group detection method based on Structured Prediction for middle density crowds. First of all, we extend the features of pairwise trajectories with velocity and orientation to obtain more valid information. Then, a fully-connected social network is maintained to improve time efficiency significantly. Finally, we adopt the adaptive-sampling BCFW algorithm to learn the mapping from trajectories to groups. Comparing with current state-of-the-art methods, our experiments demonstrate the group detection capacity on precision and time efficiency.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128829890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-08-01DOI: 10.1109/CASE48305.2020.9217025
Tristan Schnell, C. Plasberg, Lennart Puck, Timothee Buettner, Christian Eichmann, G. Heppner, A. Rönnau, R. Dillmann
Autonomous robots in complex environments are usually forced to act very conservatively, greatly limiting their potential. Taking more risky actions confidently requires the robot to have a deep understanding of its abilities, especially in its current state. The foundation for such a self-awareness is knowledge about current damages and the stress the different components of the robot are under. While the skills of a robot can be modeled in advance, the potential errors that might occur cannot easily be predicted exhaustively. Due to this, the robot is required to notice unforeseen changes in itself and judge their severity. This work presents a solution for this in the form of a Gaussian Mixture Model based framework for anomaly detection. The model requires only training data for a healthy robot, with no samples needed for expected problems and is able to correctly notice, localize and quantity various introduced damages and impairments. Transfer to new robots requires a user to only specify available sensor data for the robot’s different components. It was implemented and tested on two different robots sharing no hardware, with different problems introduced into both systems. This approach lays the foundation for a general framework for adaptive self-aware robot decision making and planning.
{"title":"Robot Health Estimation through Unsupervised Anomaly Detection using Gaussian Mixture Models","authors":"Tristan Schnell, C. Plasberg, Lennart Puck, Timothee Buettner, Christian Eichmann, G. Heppner, A. Rönnau, R. Dillmann","doi":"10.1109/CASE48305.2020.9217025","DOIUrl":"https://doi.org/10.1109/CASE48305.2020.9217025","url":null,"abstract":"Autonomous robots in complex environments are usually forced to act very conservatively, greatly limiting their potential. Taking more risky actions confidently requires the robot to have a deep understanding of its abilities, especially in its current state. The foundation for such a self-awareness is knowledge about current damages and the stress the different components of the robot are under. While the skills of a robot can be modeled in advance, the potential errors that might occur cannot easily be predicted exhaustively. Due to this, the robot is required to notice unforeseen changes in itself and judge their severity. This work presents a solution for this in the form of a Gaussian Mixture Model based framework for anomaly detection. The model requires only training data for a healthy robot, with no samples needed for expected problems and is able to correctly notice, localize and quantity various introduced damages and impairments. Transfer to new robots requires a user to only specify available sensor data for the robot’s different components. It was implemented and tested on two different robots sharing no hardware, with different problems introduced into both systems. This approach lays the foundation for a general framework for adaptive self-aware robot decision making and planning.","PeriodicalId":212181,"journal":{"name":"2020 IEEE 16th International Conference on Automation Science and Engineering (CASE)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128357043","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：