An assembly robot needs to be capable of executing an assembly task robustly under various uncertainties.To attain this goal,we use a task sequence tree model originally proposed for manual assembly.This model regards...An assembly robot needs to be capable of executing an assembly task robustly under various uncertainties.To attain this goal,we use a task sequence tree model originally proposed for manual assembly.This model regards an assembly task under uncertainties as a transformation of the contact state concept.The concept may contain several contact states with probabilities but these are transformed through a series of task elements into the contact state concept having only the goal state at the end.The transformed contact state concept can be classified according to the terminal condition of each task element.Thus,the whole assembly task can be designed as a tree-shaped contingent strategy called a task sequence tree.This paper proposes a systematic approach for reconfiguring a task sequence tree model for application to a robotic assembly task.In addition,by taking a 2D peg-in-hole insertion task to be performed by a robot equipped with a force sensor as an example,we confirm that the proposed approach can provide a robust motion strategy for the task and that the robot can actually execute the task robustly under bounded uncertainty according to the strategy.展开更多
在多无人机协同突防过程中,针对无人机需完成目标分配与探干侦动作选择的任务决策存在模型构建难,对模型求解方法复杂度高的问题,提出一种蒙特卡罗树搜索(Monte Carlo tree search,MCTS)的改进方法实现多无人机多任务联合决策。首先,考...在多无人机协同突防过程中,针对无人机需完成目标分配与探干侦动作选择的任务决策存在模型构建难,对模型求解方法复杂度高的问题,提出一种蒙特卡罗树搜索(Monte Carlo tree search,MCTS)的改进方法实现多无人机多任务联合决策。首先,考虑无人机与雷达对抗中的角度、距离等态势要素,以及当前无人机动作执行成功概率和雷达状态有效概率,构建多无人机目标分配与探干侦动作统一决策数学模型。其次,提出搜索次数自适应调整的改进MCTS算法对模型求解,实现大规模解空间在线快速寻优。仿真结果表明,改进算法使多雷达系统对无人机的威胁程度下降约16.8%,相比多臂赌博机算法效果提升约5.08%,决策时间约0.23 s,比传统MCTS缩短约45.7%,有助于提升无人机战场生存率。展开更多
面向插电式混合动力乘用车(Plug-in hybrid electric passenger vehicles,PHEV)能耗表征,构建同时预测油耗(Fuel consumption,FC)与等效电耗(Equivalent power consumption,EPC)的联合建模方案.以美国能源署官方公开数据的PHEV子集为样...面向插电式混合动力乘用车(Plug-in hybrid electric passenger vehicles,PHEV)能耗表征,构建同时预测油耗(Fuel consumption,FC)与等效电耗(Equivalent power consumption,EPC)的联合建模方案.以美国能源署官方公开数据的PHEV子集为样本,采用共享特征的梯度提升决策树(Gradient Boosting Decision Tree,GBDT)对两通道回归,设置单任务线性/树模型为基线,按“车辆类别×驱动形式”分层的五折交叉验证进行折外评估.引入纯电占比α,定义等效能耗(Equivalent energy consumption,EEC)并分析情景敏感性与误差传递.结果显示:折外平均绝对误差(Out-of-Fold Mean Absolute Error,OOFMAE)约0.9 MPG(miles per gallon)、3.9 EMPG(equivalent miles per gallon),R^(2)约0.97、0.93;EEC误差随α从FC主导向EPC.分层统计表明,不同车辆类别与驱动形式的误差存在差异.展开更多
The real problem in cluster of workstations is the changes in workstation power or number of workstations or dynmaic changes in the run time behavior of the application hamper the efficient use of resources. Dynamic l...The real problem in cluster of workstations is the changes in workstation power or number of workstations or dynmaic changes in the run time behavior of the application hamper the efficient use of resources. Dynamic load balancing is a technique for the parallel implementation of problems, which generate unpredictable workloads by migration work units from heavily loaded processor to lightly loaded processors at run time. This paper proposed an efficient load balancing method in which parallel tree computations depth first search (DFS) generates unpredictable, highly imbalance workloads and moves through different phases detectable at run time, where dynamic load balancing strategy is applicable in each phase running under the MPI(message passing interface) and Unix operating system on cluster of workstations parallel platform computing.展开更多
针对多视觉任务中传输成本高、解码端计算压力大的问题,提出一种自适应可伸缩视频编码(adaptive scalable video coding,ASVC)传输框架,将视频分为语义层和背景层,分别传输语义和背景信息。此外,提出一种自适应压缩算法,构建了C4.5决策...针对多视觉任务中传输成本高、解码端计算压力大的问题,提出一种自适应可伸缩视频编码(adaptive scalable video coding,ASVC)传输框架,将视频分为语义层和背景层,分别传输语义和背景信息。此外,提出一种自适应压缩算法,构建了C4.5决策树模型分析网络环境对视频进行压缩的决策判定,并对帧序列进行光流分析,在保留变化显著的帧基础上引入插值机制保持图像的平滑性。仿真结果表明,ASVC方法在不同码率环境下表现更高的识别精准率,视频质量和传输效率的显著提升。展开更多
基金Project (No.19GS0208) supported by the Grant-in-Aid for Creative Scientific Research 2007–2011 funded by the Ministry of Education,Culture,Sports,Science and Technology,Japan
文摘An assembly robot needs to be capable of executing an assembly task robustly under various uncertainties.To attain this goal,we use a task sequence tree model originally proposed for manual assembly.This model regards an assembly task under uncertainties as a transformation of the contact state concept.The concept may contain several contact states with probabilities but these are transformed through a series of task elements into the contact state concept having only the goal state at the end.The transformed contact state concept can be classified according to the terminal condition of each task element.Thus,the whole assembly task can be designed as a tree-shaped contingent strategy called a task sequence tree.This paper proposes a systematic approach for reconfiguring a task sequence tree model for application to a robotic assembly task.In addition,by taking a 2D peg-in-hole insertion task to be performed by a robot equipped with a force sensor as an example,we confirm that the proposed approach can provide a robust motion strategy for the task and that the robot can actually execute the task robustly under bounded uncertainty according to the strategy.
文摘在多无人机协同突防过程中,针对无人机需完成目标分配与探干侦动作选择的任务决策存在模型构建难,对模型求解方法复杂度高的问题,提出一种蒙特卡罗树搜索(Monte Carlo tree search,MCTS)的改进方法实现多无人机多任务联合决策。首先,考虑无人机与雷达对抗中的角度、距离等态势要素,以及当前无人机动作执行成功概率和雷达状态有效概率,构建多无人机目标分配与探干侦动作统一决策数学模型。其次,提出搜索次数自适应调整的改进MCTS算法对模型求解,实现大规模解空间在线快速寻优。仿真结果表明,改进算法使多雷达系统对无人机的威胁程度下降约16.8%,相比多臂赌博机算法效果提升约5.08%,决策时间约0.23 s,比传统MCTS缩短约45.7%,有助于提升无人机战场生存率。
文摘面向插电式混合动力乘用车(Plug-in hybrid electric passenger vehicles,PHEV)能耗表征,构建同时预测油耗(Fuel consumption,FC)与等效电耗(Equivalent power consumption,EPC)的联合建模方案.以美国能源署官方公开数据的PHEV子集为样本,采用共享特征的梯度提升决策树(Gradient Boosting Decision Tree,GBDT)对两通道回归,设置单任务线性/树模型为基线,按“车辆类别×驱动形式”分层的五折交叉验证进行折外评估.引入纯电占比α,定义等效能耗(Equivalent energy consumption,EEC)并分析情景敏感性与误差传递.结果显示:折外平均绝对误差(Out-of-Fold Mean Absolute Error,OOFMAE)约0.9 MPG(miles per gallon)、3.9 EMPG(equivalent miles per gallon),R^(2)约0.97、0.93;EEC误差随α从FC主导向EPC.分层统计表明,不同车辆类别与驱动形式的误差存在差异.
基金Natural Science Foundation of China (No.60 173 0 3 1)
文摘The real problem in cluster of workstations is the changes in workstation power or number of workstations or dynmaic changes in the run time behavior of the application hamper the efficient use of resources. Dynamic load balancing is a technique for the parallel implementation of problems, which generate unpredictable workloads by migration work units from heavily loaded processor to lightly loaded processors at run time. This paper proposed an efficient load balancing method in which parallel tree computations depth first search (DFS) generates unpredictable, highly imbalance workloads and moves through different phases detectable at run time, where dynamic load balancing strategy is applicable in each phase running under the MPI(message passing interface) and Unix operating system on cluster of workstations parallel platform computing.
文摘针对多视觉任务中传输成本高、解码端计算压力大的问题,提出一种自适应可伸缩视频编码(adaptive scalable video coding,ASVC)传输框架,将视频分为语义层和背景层,分别传输语义和背景信息。此外,提出一种自适应压缩算法,构建了C4.5决策树模型分析网络环境对视频进行压缩的决策判定,并对帧序列进行光流分析,在保留变化显著的帧基础上引入插值机制保持图像的平滑性。仿真结果表明,ASVC方法在不同码率环境下表现更高的识别精准率,视频质量和传输效率的显著提升。
