In this paper, we study the nonlinear regression problem in a network of nodes and introduce long short term memory (LSTM) based algorithms. In order to learn the parameters of the LSTM architecture in an online manne...
详细信息
ISBN:
(纸本)9781509064946
In this paper, we study the nonlinear regression problem in a network of nodes and introduce long short term memory (LSTM) based algorithms. In order to learn the parameters of the LSTM architecture in an online manner, we put. the LSTM equations into a nonlinear state space form and then introduce our distributed particle filtering (DPF) based training algorithm. Our training algorithm asymptotically achieves the optimal training performance. In our simulations, we illustrate the performance improvement achieved by the introduced algorithm with respect to the conventional methods.
The present work addresses the distributed Multi-Agent Multi-Object Tracking problem where a team of robots has to perform a distributed position estimation of multiple moving objects. In complex scenarios, where mobi...
详细信息
ISBN:
(纸本)9781479977727
The present work addresses the distributed Multi-Agent Multi-Object Tracking problem where a team of robots has to perform a distributed position estimation of multiple moving objects. In complex scenarios, where mobile robots are involved, it is crucial to disseminate reliable beliefs in order to avoid the degradation of the global estimations. To this end, distributedparticle Filters have been proven to be effective tools to model non-linear and dynamic processes in Multi-Robot Systems. We present therefore an asynchronous method for distributed particle filtering based on Multi-Clustered particlefiltering that uses a novel clustering technique to continuously keep track of a variable and unknown number of objects. A two-tiered architecture is proposed: a local estimation layer uses a particle Filter to integrate local observations of multiple objects detected in the local frame, while a global estimation layer is used to perform a distributed estimation integrating information collected from the other robots. We carried out a quantitative evaluation demonstrating how our proposed approach has significantly better robustness to perception noise when using mobile sensors rather than fixed sensors.
distributed PF (DPF) was used due to the limitation of nodes' computing capacity inferring to the target tracking in a wireless sensor network (WSN). In this paper, a novel filtering method - DPF* in WSN is propos...
详细信息
ISBN:
(纸本)9780769536545
distributed PF (DPF) was used due to the limitation of nodes' computing capacity inferring to the target tracking in a wireless sensor network (WSN). In this paper, a novel filtering method - DPF* in WSN is proposed. Instead of transferring value and weight of particles, Gaussian mixture model (GMM) is used to approximate the posteriori distribution, and only GMM parameters need to be transferred which can reduce the bandwidth and power consumption. In order to use sampling information effectively, when target moving to the next cluster head region, the GMM parameters are transfer to the next cluster head, and combine with the new local GMM parameters to compose the new GMM parameters incrementally. The proposed DPF* is compared to some other DPF for WSN target tracking, and the experimental results show that not the precision is improved.
暂无评论