Pub Date : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10011689
Xufei Yan, Renliang Chen, Shiqiang Zhu, Anhuan Xie, J. Gu
This paper investigates the optimal landing trajectory and control procedure when a helicopter undergoes autorotation due to tail rotor drive failure (TRDF), in which an optimal control methodology is proposed. First, a helicopter flight dynamics model with TRDF was developed. Then, the autorotation in TRDF was converted to be a nonlinear optimal control problem, solved by direct node collocation method and sequential quadratic programming algorithm. Finally, a model helicopter (Z11) with single main rotor and tail rotor was used to demonstrate the proposed approach. An optimal autorotation landing procedure in TRDF was determined accordingly. Results indicate that the airframe will immediately respond to the excess torque generated by the main rotor via yawing, sideslip and rolling when TRDF occurs. The pilot is recommended to shut down the engine and perform a series of critical operations to stabilize the violent yaw and roll movements. In addition, flight test data were used to validate the numerical simulations of autorotation landing. The proposed optimal control approach provides a useful tool to investigate helicopter TRDF autorotation landing procedure.
{"title":"Study of Helicopter Optimal Autorotation Landing Procedure in Tail Rotor Drive Failure","authors":"Xufei Yan, Renliang Chen, Shiqiang Zhu, Anhuan Xie, J. Gu","doi":"10.1109/ROBIO55434.2022.10011689","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011689","url":null,"abstract":"This paper investigates the optimal landing trajectory and control procedure when a helicopter undergoes autorotation due to tail rotor drive failure (TRDF), in which an optimal control methodology is proposed. First, a helicopter flight dynamics model with TRDF was developed. Then, the autorotation in TRDF was converted to be a nonlinear optimal control problem, solved by direct node collocation method and sequential quadratic programming algorithm. Finally, a model helicopter (Z11) with single main rotor and tail rotor was used to demonstrate the proposed approach. An optimal autorotation landing procedure in TRDF was determined accordingly. Results indicate that the airframe will immediately respond to the excess torque generated by the main rotor via yawing, sideslip and rolling when TRDF occurs. The pilot is recommended to shut down the engine and perform a series of critical operations to stabilize the violent yaw and roll movements. In addition, flight test data were used to validate the numerical simulations of autorotation landing. The proposed optimal control approach provides a useful tool to investigate helicopter TRDF autorotation landing procedure.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115762886","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 : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10011855
Shixiong Li, Zhilei Yan, Biao Hu
Robotic systems have now become very complex because they need to integrate many real-time computing tasks like environment perception, path planning and manipulator control. Due to the limitation of computing capability in robots, a cloud computing platform is often used to provide the service to robots. How to make full use of hardware resources in cloud computing to run robots' tasks in real time become a very important problem. In this paper, we develop a particle swarm optimization approach to optimize the system's resource allocation and task assignment for the aim of minimizing system's power consumption. Specifically, in a cloud computing-based robotic system, we divide hardware resource into many identical parts, and use the virtual machine technology to create some isolated virtual machines that occupy a certain amount of hardware resource. The particle swarm optimization approach is developed to allocate hardware resource and assign real-time computing tasks to these virtual machines. The optimization aims to minimize the system's power consumption because a low-power system can reduce the service cost. We propose using 2-segment code to encode the schedule into a particle, and propose a 2-step heuristic to initialize particles. Simulation experiments show that 2-segment code makes PSO applicable to solve our problem, and 2-step heuristic improves the search efficiency and solution quality.
{"title":"A PSO-based Resource Allocation and Task Assignment Approach for Real-Time Cloud Computing-based Robotic Systems","authors":"Shixiong Li, Zhilei Yan, Biao Hu","doi":"10.1109/ROBIO55434.2022.10011855","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011855","url":null,"abstract":"Robotic systems have now become very complex because they need to integrate many real-time computing tasks like environment perception, path planning and manipulator control. Due to the limitation of computing capability in robots, a cloud computing platform is often used to provide the service to robots. How to make full use of hardware resources in cloud computing to run robots' tasks in real time become a very important problem. In this paper, we develop a particle swarm optimization approach to optimize the system's resource allocation and task assignment for the aim of minimizing system's power consumption. Specifically, in a cloud computing-based robotic system, we divide hardware resource into many identical parts, and use the virtual machine technology to create some isolated virtual machines that occupy a certain amount of hardware resource. The particle swarm optimization approach is developed to allocate hardware resource and assign real-time computing tasks to these virtual machines. The optimization aims to minimize the system's power consumption because a low-power system can reduce the service cost. We propose using 2-segment code to encode the schedule into a particle, and propose a 2-step heuristic to initialize particles. Simulation experiments show that 2-segment code makes PSO applicable to solve our problem, and 2-step heuristic improves the search efficiency and solution quality.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114468543","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}
The Flapping-Wing Micro Air Vehicle, imitating the flight of birds or insects, is a new type of aircraft with broad application value. Since its flying principle is quite different from that of traditional aircraft, the efficient design of the bionic wing is the core of whether the flapping-wing micro air vehicle can have good flight performance. To achieve better flight performance, this paper analyzes the flapping mechanism based on aerodynamics, optimizes the design of the flapping mechanism of the flying robot, and improves the energy utilization rate and lift of the air vehicle. With the goal of optimizing the bionic wing, many bionic wings are completed, controlling the variables except the leading edge, to carry out the lift test, hoping to obtain the relationship between the leading-edge angle and the lift. To finish the experiment with accurate results, this paper designs a micro-bionic flapping-wing optimization design platform to verify the different wings' lift effects. Through the experiments done on the platform, a wing design scheme is found, which has the best lift performance under current conditions, and the platform can carry out flap experiments conveniently, efficiently, and accurately. Besides, the experimental result clearly shows the trend of the wing lift with the leading-edge angle, which provides an important reference for further optimization design, and has a high Reference value. Future research will concentrate on the other variables of the bionic wing design to achieve more lift.
{"title":"Design and Optimization of Bionic Wings Based on Leading-edge Angle for Flapping-Wing Micro Air Vehicle","authors":"Yuchen Xia, Huichao Deng, Kai Hu, Lili Yang, Shengjie Xiao, Xilun Ding, Zhaolu Xiong","doi":"10.1109/ROBIO55434.2022.10011821","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011821","url":null,"abstract":"The Flapping-Wing Micro Air Vehicle, imitating the flight of birds or insects, is a new type of aircraft with broad application value. Since its flying principle is quite different from that of traditional aircraft, the efficient design of the bionic wing is the core of whether the flapping-wing micro air vehicle can have good flight performance. To achieve better flight performance, this paper analyzes the flapping mechanism based on aerodynamics, optimizes the design of the flapping mechanism of the flying robot, and improves the energy utilization rate and lift of the air vehicle. With the goal of optimizing the bionic wing, many bionic wings are completed, controlling the variables except the leading edge, to carry out the lift test, hoping to obtain the relationship between the leading-edge angle and the lift. To finish the experiment with accurate results, this paper designs a micro-bionic flapping-wing optimization design platform to verify the different wings' lift effects. Through the experiments done on the platform, a wing design scheme is found, which has the best lift performance under current conditions, and the platform can carry out flap experiments conveniently, efficiently, and accurately. Besides, the experimental result clearly shows the trend of the wing lift with the leading-edge angle, which provides an important reference for further optimization design, and has a high Reference value. Future research will concentrate on the other variables of the bionic wing design to achieve more lift.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117049422","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}
Under the circumstance of COVID-19 epidemic spread, global medical resources are in serious shortage. As a common way of care for respiratory diseases, although back-slap sputum excretion can be used for the care of lung diseases, but it requires the cooperation of multiple medical staff, and lead to inefficient care. This paper designed a method of the human” s back feature recognition based on YOLOv5, and built a new type of intelligent robot for back-slap sputum excretion on this basis, which can assist care staff to complete the back-slap sputum excretion care for patients, and reduce the labor intensity of staff and the risk of cross infection.
{"title":"Design of an Intelligent Robot for Back-Slap Sputum Excretion Based on Back Feature Recognition","authors":"Diansheng Chen, Yue Pan, Yuanhai Huang, Min Wang, Renren Bao, Chunxia Tang","doi":"10.1109/ROBIO55434.2022.10011775","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011775","url":null,"abstract":"Under the circumstance of COVID-19 epidemic spread, global medical resources are in serious shortage. As a common way of care for respiratory diseases, although back-slap sputum excretion can be used for the care of lung diseases, but it requires the cooperation of multiple medical staff, and lead to inefficient care. This paper designed a method of the human” s back feature recognition based on YOLOv5, and built a new type of intelligent robot for back-slap sputum excretion on this basis, which can assist care staff to complete the back-slap sputum excretion care for patients, and reduce the labor intensity of staff and the risk of cross infection.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122095044","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}
An adaptive sensorized robot foot that can adapt to uneven and rough terrains and have contact state sensing capability is vital for legged locomotion in exploration missions. Inspired by quadruped animals, we present an adaptive planar foot with large contact area to reduce sinkage, two degrees of freedom (2-DOF) compliant ankle joint for better terrain adaptation, in-foot multi-modal sensing capability. We propose customized ankle design with encoders instead of IMU for ankle joint angle measurement, in order to mitigate the noise injected by large impact forces. The foot also provides an accurate estimated force according to foot-shank relative orientation and the 6-axis force/torque sensor data. We develop a control algorithm for a quadruped robot equipped with our proposed robot feet and test its performance through simulation. Real-world experiments including foot orientation estimation, swing sensitivity, and force perception verify that our robot foot is suitable for dynamic quadruped locomotion.
{"title":"Adaptive Planar Foot with Compliant Ankle Joint and Multi-modal Sensing for Quadruped Robots","authors":"Guowei Shi, Chen Yao, Wenhui Wang, Zheng Zhu, Zhenzhong Jia","doi":"10.1109/ROBIO55434.2022.10011981","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011981","url":null,"abstract":"An adaptive sensorized robot foot that can adapt to uneven and rough terrains and have contact state sensing capability is vital for legged locomotion in exploration missions. Inspired by quadruped animals, we present an adaptive planar foot with large contact area to reduce sinkage, two degrees of freedom (2-DOF) compliant ankle joint for better terrain adaptation, in-foot multi-modal sensing capability. We propose customized ankle design with encoders instead of IMU for ankle joint angle measurement, in order to mitigate the noise injected by large impact forces. The foot also provides an accurate estimated force according to foot-shank relative orientation and the 6-axis force/torque sensor data. We develop a control algorithm for a quadruped robot equipped with our proposed robot feet and test its performance through simulation. Real-world experiments including foot orientation estimation, swing sensitivity, and force perception verify that our robot foot is suitable for dynamic quadruped locomotion.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125802775","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 : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10012009
Jianyin Tang, Hong Tao, Xinhan Zhuang, Yang Cheng, Hang Xiao, Kun Xu, Xilun Ding
This paper describes a robotic arm that allows for safe human-robot interaction. The robotic arm is essentially a cable-driven robotic arm with passive variable stiffness features; the cable-driven section of the arm maintains high stiffness and strength by utilizing a unique lightweight tension amplification system. The stiffness of the joints, which is intimately related to the performance of motion control, is quadratically amplified. We propose a 1-DOF and 3-DOF joint mechanism using the tension amplification method, which we combine to construct the robotic arm's elbow and wrist. Modular joints with variable stiffness are used in the 3-DOF shoulder. The passive compliant parts of the variable stiffness modular joint with progressive stiffness could make the actuator inherently soft at low contact torque levels and significantly stiffer at higher interaction torque levels, resolving common design trade-offs in linear series elastic actuators. To assess the workspace of the robotic arm, a kinematic model of the 7-DOF robotic arm is built. Finally, a cable-driven robotic arm prototype with variable stiffness joint module is developed.
{"title":"Design and analysis of cable-driven robotic arm with variable stiffness modular joint","authors":"Jianyin Tang, Hong Tao, Xinhan Zhuang, Yang Cheng, Hang Xiao, Kun Xu, Xilun Ding","doi":"10.1109/ROBIO55434.2022.10012009","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10012009","url":null,"abstract":"This paper describes a robotic arm that allows for safe human-robot interaction. The robotic arm is essentially a cable-driven robotic arm with passive variable stiffness features; the cable-driven section of the arm maintains high stiffness and strength by utilizing a unique lightweight tension amplification system. The stiffness of the joints, which is intimately related to the performance of motion control, is quadratically amplified. We propose a 1-DOF and 3-DOF joint mechanism using the tension amplification method, which we combine to construct the robotic arm's elbow and wrist. Modular joints with variable stiffness are used in the 3-DOF shoulder. The passive compliant parts of the variable stiffness modular joint with progressive stiffness could make the actuator inherently soft at low contact torque levels and significantly stiffer at higher interaction torque levels, resolving common design trade-offs in linear series elastic actuators. To assess the workspace of the robotic arm, a kinematic model of the 7-DOF robotic arm is built. Finally, a cable-driven robotic arm prototype with variable stiffness joint module is developed.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125883689","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 : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10011675
Zhengyang Li, Qingsong Xu
This paper presents the design and development of a magnetic-actuated multi-segment robotic catheter (MMR-C), which is driven by an external permanent mobile magnet system (EPMM). The structure of the multi-segment catheter is devised by the integration of variable outer diameter of each segment and multiple opposite polarization magnets. The system is designed for the target intervention application scenario of minimally invasive surgery in a multi-branch vascular cavity. The kinematic model of MMRC is derived based on the Cosserat rod method. The control scheme for the EPMM and the MMRC is introduced. The trajectories of the EPMM in Cartesian space are realized by the combination of dynamic movement primitives and Gaussian Mixture Regression. More over, the effectiveness of the proposed robotic system has been verified by conducting several experimental studies. The system performance is demonstrated by the carried out ring steering test and in-vitro vascular phantom intervention test.
{"title":"Design and Steering Control of a New Magnetic-Actuated Multi-Segment Robotic Catheter","authors":"Zhengyang Li, Qingsong Xu","doi":"10.1109/ROBIO55434.2022.10011675","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011675","url":null,"abstract":"This paper presents the design and development of a magnetic-actuated multi-segment robotic catheter (MMR-C), which is driven by an external permanent mobile magnet system (EPMM). The structure of the multi-segment catheter is devised by the integration of variable outer diameter of each segment and multiple opposite polarization magnets. The system is designed for the target intervention application scenario of minimally invasive surgery in a multi-branch vascular cavity. The kinematic model of MMRC is derived based on the Cosserat rod method. The control scheme for the EPMM and the MMRC is introduced. The trajectories of the EPMM in Cartesian space are realized by the combination of dynamic movement primitives and Gaussian Mixture Regression. More over, the effectiveness of the proposed robotic system has been verified by conducting several experimental studies. The system performance is demonstrated by the carried out ring steering test and in-vitro vascular phantom intervention test.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125933280","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 : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10011712
Miaomiao Wang, Zhendong Sun
A switched system is composed by continuous dynamics, logic dynamics, and the interaction between them. For the general class of controllable and reversible discrete-time switched linear control systems, this work presents a computational design scheme that could optimize the transient response in terms of state overshoot and settling time. The scheme is based on a two-step optimization: the first is to design the deadbeat feedback gain matrices by exhaust searching the controllability switching realizations, and the second is to optimize the feedback gain matrices by random perturbations. A numerical example is presented to show the effectiveness of the proposed design scheme.
{"title":"Improving Transient Response of a Class of Switched Linear Systems*","authors":"Miaomiao Wang, Zhendong Sun","doi":"10.1109/ROBIO55434.2022.10011712","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011712","url":null,"abstract":"A switched system is composed by continuous dynamics, logic dynamics, and the interaction between them. For the general class of controllable and reversible discrete-time switched linear control systems, this work presents a computational design scheme that could optimize the transient response in terms of state overshoot and settling time. The scheme is based on a two-step optimization: the first is to design the deadbeat feedback gain matrices by exhaust searching the controllability switching realizations, and the second is to optimize the feedback gain matrices by random perturbations. A numerical example is presented to show the effectiveness of the proposed design scheme.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129415661","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 : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10011650
Jiangqin Deng, Ziqing Li, Yang Zheng, Guoying Gu
Hyper-redundant manipulators have redundant degrees of freedom, bringing additional difficulties to the motion planning. Generally, existing motion planning methods generate paths without considering joint angle limitation of hyper-redundant manipulators. And the generated paths are discretized, leading to discrete errors. In this paper, we present an autonomous motion planner to generate paths, which can be followed by hyper-redundant manipulators with analytical solutions. Starting from the ending pose in the workspace, the rapidly exploring random tree can expand to multiple entrances with limited curvature of arc segments, which ensures that the joint angle limitation is satisfied. Meanwhile, the generated path consists of arc segments, which makes the generated paths can be followed with analytical solutions. Several simulations are conducted to demonstrate the aforementioned advantages. For further validation of the planner's effectiveness, a hyper-redundant manipulator system is used to follow the generated path with follow-the-leader motion.
{"title":"An RRT-Based Motion Planning Method for Hyper-Redundant Manipulators in Confined Spaces","authors":"Jiangqin Deng, Ziqing Li, Yang Zheng, Guoying Gu","doi":"10.1109/ROBIO55434.2022.10011650","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011650","url":null,"abstract":"Hyper-redundant manipulators have redundant degrees of freedom, bringing additional difficulties to the motion planning. Generally, existing motion planning methods generate paths without considering joint angle limitation of hyper-redundant manipulators. And the generated paths are discretized, leading to discrete errors. In this paper, we present an autonomous motion planner to generate paths, which can be followed by hyper-redundant manipulators with analytical solutions. Starting from the ending pose in the workspace, the rapidly exploring random tree can expand to multiple entrances with limited curvature of arc segments, which ensures that the joint angle limitation is satisfied. Meanwhile, the generated path consists of arc segments, which makes the generated paths can be followed with analytical solutions. Several simulations are conducted to demonstrate the aforementioned advantages. For further validation of the planner's effectiveness, a hyper-redundant manipulator system is used to follow the generated path with follow-the-leader motion.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128590485","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 : 2022-12-05DOI: 10.1109/ROBIO55434.2022.10011977
Yue Gao, Qiyue Yang, Huajian Wu, Mingdong Sun
Nonlinear least-squares problems (NLS) are pop-ular in engineering and scientific fields. Traditional optimization methods such as Newton's method and Gaussian-Newton method (GN) suffer from the sensibility to initial values and the high computational complexity. In this paper, we propose LS-DDPG, a robust optimization method utilizing deep rein-forcement learning algorithms to solve nonlinear least-squares problems. The experiment results on synthetic data demonstrate that the proposed method outperforms Newton's method in terms of computation cost, convergence speed and initial values sensibility. In addition, LS-DDPG is utilized on model predictive control (MPC) problems for trajectory planning and tracking tasks in self-driving with longer prediction horizon and higher accuracy than baseline methods.
{"title":"Learning to Solve Nonlinear Optimization Problem with Deep Reinforcement Learning","authors":"Yue Gao, Qiyue Yang, Huajian Wu, Mingdong Sun","doi":"10.1109/ROBIO55434.2022.10011977","DOIUrl":"https://doi.org/10.1109/ROBIO55434.2022.10011977","url":null,"abstract":"Nonlinear least-squares problems (NLS) are pop-ular in engineering and scientific fields. Traditional optimization methods such as Newton's method and Gaussian-Newton method (GN) suffer from the sensibility to initial values and the high computational complexity. In this paper, we propose LS-DDPG, a robust optimization method utilizing deep rein-forcement learning algorithms to solve nonlinear least-squares problems. The experiment results on synthetic data demonstrate that the proposed method outperforms Newton's method in terms of computation cost, convergence speed and initial values sensibility. In addition, LS-DDPG is utilized on model predictive control (MPC) problems for trajectory planning and tracking tasks in self-driving with longer prediction horizon and higher accuracy than baseline methods.","PeriodicalId":151112,"journal":{"name":"2022 IEEE International Conference on Robotics and Biomimetics (ROBIO)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127493272","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：