Biomimetic grasping is crucial for robots to interact with the environment and perform complex tasks,making it a key focus in robotics and embodied intelligence.However,achieving human-level finger coordination and fo...Biomimetic grasping is crucial for robots to interact with the environment and perform complex tasks,making it a key focus in robotics and embodied intelligence.However,achieving human-level finger coordination and force control remains challenging due to the need for multimodal perception,including visual,kinesthetic,and tactile feedback.Although some recent approaches have demonstrated remarkable performance in grasping diverse objects,they often rely on expensive tactile sensors or are restricted to rigid objects.To address these challenges,we introduce SoftGrasp,a novel multimodal imitation learning approach for adaptive,multi-stage grasping of objects with varying sizes,shapes,and hardness.First,we develop an immersive demonstration platform with force feedback to collect rich,human-like grasping datasets.Inspired by human proprioceptive manipulation,this platform gathers multimodal signals,including visual images,robot finger joint angles,and joint torques,during demonstrations.Next,we utilize a multi-head attention mechanism to align and integrate multimodal features,dynamically allocating attention to ensure comprehensive learning.On this basis,we design a behavior cloning method based on an angle-torque loss function,enabling multimodal imitation learning.Finally,we validate SoftGrasp in extensive experiments across various scenarios,demonstrating its ability to adaptively adjust joint forces and finger angles based on real-time inputs.These capabilities result in a 98%success rate in real-world experiments,achieving dexterous and stable grasping.Source code and demonstration videos are available at https://github.com/nubot-nudt/SoftGrasp.展开更多
Inspired by the morphology of human fingers,this paper proposes an underactuated rigid-soft coupled robotic gripper whose finger is designed as the combination of a rigid skeleton and a soft tissue.Different from the ...Inspired by the morphology of human fingers,this paper proposes an underactuated rigid-soft coupled robotic gripper whose finger is designed as the combination of a rigid skeleton and a soft tissue.Different from the current grippers who have multi-point contact or line contact with the target objects,the proposed robotic gripper enables surface contact and leads to flexible grasping and robust holding.The actuated mechanism,which is the palm of proposed gripper,is optimized for excellent operability based on a mathematical model.Soft material selection and rigid skeleton structure of fingers are then analyzed through a series of dynamic simulations by RecurDyn and Adams.After above design process including topology analysis,actuated mechanism optimization,soft material selection and rigid skeleton analysis,the rigid-soft coupled robotic gripper is fabricated via 3D printing.Finally,the grasping and holding capabilities are validated by experiments testing the stiffness of a single finger and the impact resistance of the gripper.Experimental results show that the proposed rigid-soft coupled robotic gripper can adapt to objects with different properties(shape,size,weight and softness)and hold them steadily.It confirms the feasibility of the design procedure,as well as the compliant and dexterous grasping capabilities of proposed rigid-soft coupled gripper.展开更多
Friction plays a critical role in dexterous robotic manipulation.However,realizing friction sensing remains a challenge due to the difficulty in designing sensing structures to decouple multi-axial forces.Inspired by ...Friction plays a critical role in dexterous robotic manipulation.However,realizing friction sensing remains a challenge due to the difficulty in designing sensing structures to decouple multi-axial forces.Inspired by the topological mechanics of knots,we construct optical fiber knot(OFN)sensors for slip detection and friction measurement.By introducing localized self-contacts along the fiber,the knot structure enables anisotropic responses to normal and frictional forces.By employing OFNs and a change point detection algorithm,we demonstrate adaptive robotic grasping of slipping cups.We further develop a robotic finger that can measure tri-axial forces via a centrosymmetric architecture composed of five OFNs.Such a tactile finger allows a robotic hand to manipulate human tools dexterously.This work could provide a straightforward and cost-effective strategy for promoting adaptive grasping,dexterous manipulation,and human-robot interaction with tactile sensing.展开更多
Circuit partitioning plays a crucial role in very large-scale integrated circuit (VLSI) physical design automation. With current trends, partitioning with multiple objectives which includes cutsize, area, delay, and p...Circuit partitioning plays a crucial role in very large-scale integrated circuit (VLSI) physical design automation. With current trends, partitioning with multiple objectives which includes cutsize, area, delay, and power obtains much concentration. In this paper, a multi-objective greedy randomized adaptive search procedure (GRASP) is presented for simultaneous cutsize and circuit delay minimization. Each objective is assigned a preference or weight to direct the search procedure and generate a variety of efficient solutions by changing the preference. To get a good initial partition with minimal cutsize and circuit delay, the gain of each module in a circuit is computed by considering both signal nets and circuit delay. The performance of the proposed algorithm is evaluated on a standard set of partitioning benchmark. The experimental results show that the proposed algorithm can generate a set of Pareto optimal solutions and is efficient for tackling multi-objective circuit partitioning.展开更多
基金supported by the Innovation Science Foundation of National University of Defense Technology,China(24-ZZCX-GZZ-11)the National Science Foundation of China(62373201).
文摘Biomimetic grasping is crucial for robots to interact with the environment and perform complex tasks,making it a key focus in robotics and embodied intelligence.However,achieving human-level finger coordination and force control remains challenging due to the need for multimodal perception,including visual,kinesthetic,and tactile feedback.Although some recent approaches have demonstrated remarkable performance in grasping diverse objects,they often rely on expensive tactile sensors or are restricted to rigid objects.To address these challenges,we introduce SoftGrasp,a novel multimodal imitation learning approach for adaptive,multi-stage grasping of objects with varying sizes,shapes,and hardness.First,we develop an immersive demonstration platform with force feedback to collect rich,human-like grasping datasets.Inspired by human proprioceptive manipulation,this platform gathers multimodal signals,including visual images,robot finger joint angles,and joint torques,during demonstrations.Next,we utilize a multi-head attention mechanism to align and integrate multimodal features,dynamically allocating attention to ensure comprehensive learning.On this basis,we design a behavior cloning method based on an angle-torque loss function,enabling multimodal imitation learning.Finally,we validate SoftGrasp in extensive experiments across various scenarios,demonstrating its ability to adaptively adjust joint forces and finger angles based on real-time inputs.These capabilities result in a 98%success rate in real-world experiments,achieving dexterous and stable grasping.Source code and demonstration videos are available at https://github.com/nubot-nudt/SoftGrasp.
基金This was supported in part by the National Natural Science Foundation of China under Grant 52275027,52275028 and 52205028in part by the Tianjin Science and Technology Planning Project under Grant 20201193.
文摘Inspired by the morphology of human fingers,this paper proposes an underactuated rigid-soft coupled robotic gripper whose finger is designed as the combination of a rigid skeleton and a soft tissue.Different from the current grippers who have multi-point contact or line contact with the target objects,the proposed robotic gripper enables surface contact and leads to flexible grasping and robust holding.The actuated mechanism,which is the palm of proposed gripper,is optimized for excellent operability based on a mathematical model.Soft material selection and rigid skeleton structure of fingers are then analyzed through a series of dynamic simulations by RecurDyn and Adams.After above design process including topology analysis,actuated mechanism optimization,soft material selection and rigid skeleton analysis,the rigid-soft coupled robotic gripper is fabricated via 3D printing.Finally,the grasping and holding capabilities are validated by experiments testing the stiffness of a single finger and the impact resistance of the gripper.Experimental results show that the proposed rigid-soft coupled robotic gripper can adapt to objects with different properties(shape,size,weight and softness)and hold them steadily.It confirms the feasibility of the design procedure,as well as the compliant and dexterous grasping capabilities of proposed rigid-soft coupled gripper.
基金grateful for financial supports from National Natural Science Foundation of China(61975173)China Postdoctoral Science Foundation(2022M722907,2022M722909)+2 种基金Zhejiang Provincial Natural Science Foundation of China(LQ23F010015)Key Research and Development Project of Zhejiang Province(2021C05003)Major Scientific Research Project of Zhejiang Lab(2019MC0AD01).
文摘Friction plays a critical role in dexterous robotic manipulation.However,realizing friction sensing remains a challenge due to the difficulty in designing sensing structures to decouple multi-axial forces.Inspired by the topological mechanics of knots,we construct optical fiber knot(OFN)sensors for slip detection and friction measurement.By introducing localized self-contacts along the fiber,the knot structure enables anisotropic responses to normal and frictional forces.By employing OFNs and a change point detection algorithm,we demonstrate adaptive robotic grasping of slipping cups.We further develop a robotic finger that can measure tri-axial forces via a centrosymmetric architecture composed of five OFNs.Such a tactile finger allows a robotic hand to manipulate human tools dexterously.This work could provide a straightforward and cost-effective strategy for promoting adaptive grasping,dexterous manipulation,and human-robot interaction with tactile sensing.
基金National Natural Science Foudation of China (No. 61070020 )Research Foundation for Doctoral Program of Ministry of Education,China (No. 20093514110004)Foundations of Education Department of Fujian Province,China (No. JA10284,No. JB07283)
文摘Circuit partitioning plays a crucial role in very large-scale integrated circuit (VLSI) physical design automation. With current trends, partitioning with multiple objectives which includes cutsize, area, delay, and power obtains much concentration. In this paper, a multi-objective greedy randomized adaptive search procedure (GRASP) is presented for simultaneous cutsize and circuit delay minimization. Each objective is assigned a preference or weight to direct the search procedure and generate a variety of efficient solutions by changing the preference. To get a good initial partition with minimal cutsize and circuit delay, the gain of each module in a circuit is computed by considering both signal nets and circuit delay. The performance of the proposed algorithm is evaluated on a standard set of partitioning benchmark. The experimental results show that the proposed algorithm can generate a set of Pareto optimal solutions and is efficient for tackling multi-objective circuit partitioning.
