一种自适应免疫优化的无迹粒子滤波器

针对无迹粒子滤波（UPF）在较偏观测时的退化现象及重采样带来的粒子枯竭问题,提出一种自适应免疫优化的无迹粒子滤波算法（AIO-UPF）。该算法在重采样过程中,利用免疫算法在亲和度与浓度调节机制下的全局寻优能力和多样性特征,通过引入自适应阈值因子δ的Metropolis准则,使得粒子集能够有效地分布于高似然区域,提高了粒子的多样性和有效性,从而较好地抑制了在较偏观测时的粒子退化问题。仿真结果表明,AIO-UPF的性能优于传统UPF及标准粒子滤波,在状态估计精度上比传统UPF提高了27%左右。     

基于引力场优化的Unscented FastSLAM2.0算法

为了改进Unscented Fast SLAM2.0算法重采样过程中的"粒子退化"和"粒子贫化"问题,本文提出了一种基于引力场优化的Unscented Fast SLAM2.0算法.首先采用Unscented粒子滤波器替代扩展卡尔曼滤波估计移动机器人路径后验概率,然后采用扩展卡尔曼滤波器对环境进行估计更新,最后用引力场优化思想优化重采样过程:在重采样中每个采样粒子近似成宇宙灰尘,通过引力场的移动因子产生作用驱动粒子集更快朝着真实的机器人位姿状态逼近,改善粒子退化问题:通过自转因子的自转作用,避免粒子过分集中,保障了粒子多样性.实验结果表明了该算法的有效性.     

一种基于遗传算法的FastSLAM2.0算法

FastSLAM2.0算法的重采样过程会带来“粒子耗尽”问题,为了改进算法的性能、提高估计精度,将FastSLAM2.0算法与遗传算法相结合,提出了一种解决SLAM问题的方法——遗传快速SLAM算法．针对FastSLAM2.0算法的特点,设计了一种改进的遗传算法来兼顾粒子权值和粒子集的多样性．遗传快速SLAM算法采用unscented粒子滤波器估计机器人的路径,地图估计则采用扩展卡尔曼滤波器．采用SLAM领域的标准数据集“carparkdataset”对提出的算法进行了验证,实验结果表明遗传快速SLAM算法在估计精度和一致性方面都具有较好的性能,并且算法的计算复杂度能满足实时性要求．     

一种基于SR-UKF的FastSLAM算法

标准FastSLAM算法存在着粒子集退化和线性化误差累积的缺陷。针对上述问题,提出了基于平方根无迹卡尔曼滤波(SR-UKF)的FastSLAM算法。SR-UKF选取一组能够代表状态向量统计特性的代表点带入非线性函数处理后重新构建出新的统计特性;使用SR-UFK取代EKF来估计每个粒子的后验位姿提议分布,可以提高粒子采样精度,减缓粒子集的退化;同时SR-UKF可以确保协方差矩阵的非负定,保证了SLAM算法的稳定性。仿真实验结果表明,基于SR-UKF的FastSLAM算法在估计精度和鲁棒性两方面均优于FastSLAM 2.0算法。     

基于Dirichlet过程非参贝叶斯学习的高斯箱粒子滤波快速 SLAM算法

针对常规FastSLAM算法需要大量粒子创建地图以及粒子退化而导致计算复杂度高、难以提高估计精度等问题,提出了一种基于Dirichlet过程非参贝叶斯学习的高斯箱粒子滤波快速SLAM(同步定位与地图构建)算法.首先,改进了箱粒子滤波中以箱粒子为支撑集的均匀概率密度函数,采用高斯概率密度函数进行贝叶斯滤波,提高了估计的精度.在此基础上将Dirichlet过程非参贝叶斯学习应用于高斯箱粒子的重采样,既保证了有效箱粒子数,又能让箱粒子集中在高似然区域,降低了采样枯竭的影响.然后,利用基于Dirichlet过程非参贝叶斯学习的高斯箱粒子滤波进行机器人位姿估计,可有效降低地图创建所需的粒子数,并提高定位精度和实时性.进而采用无迹卡尔曼滤波更新地图特征,以提高地图创建的一致性.仿真结果和轮腿复合机器人实地实验结果验证了本文方法的可行性和有效性.     

基于粒子群优化的Unscented粒子滤波算法

针对Unscented粒子滤波(UPF)算法中的粒子退化及重采样引起的粒子枯竭等问题,利用粒子群优化算法使粒子通过比较其当前值与最优粒子的适应度值调整自身速度,向高似然域移动,寻找最优位置,并对重采样过程进行优化,以缓解粒子的退化及枯竭问题。实验结果证明,该算法提高了UPF算法的状态估计精度。     

FastSLAM算法的仿生优化改进研究

针对机器人导航标准的快速同步定位与地图构建算法(FastSLAM)在重采样过程中存在采样粒子集的贫化以及粒子多样性的缺失导致机器人的定位与建图的精度下降的问题,提出一种基于改进的蝴蝶算法来优化FastSLAM中的粒子滤波部分。改进的算法将机器人的最新时刻的观测和状态信息融入到蝴蝶算法的香味公式中,并在蝴蝶位置更新的过程加入自适应香味半径和自适应蝴蝶飞行调整步长因子,来减少算法的运算时间以及提高预测精度,同时引入偏差修正指数加权算法对粒子的权值进行优化组合,对组合后部分不稳定的粒子进行分布重采样,保证粒子的多样性。通过仿真验证了该算法在估计精度与稳定性方面优于FastSLAM,因此在移动机器人运动模型的定位与建图中具有较高的定位精度与稳定性。     

基于精确稀疏扩展信息滤波的粒子滤波SLAM算法研究

传统粒子滤波算法的单次迭代过程以及小权值粒子在重采样中被删除都使得机器人位姿的历史信息不能充分利用,因而会出现粒子的退化现象,从而导致滤波算法的估计精度较低。本文提出基于精确稀疏扩展信息滤波的粒子滤波SLAM算法,利用精确稀疏扩展信息滤波的信息矩阵反映机器人位姿相对变化的同时,也对应于状态后验概率的条件概率的性质,应用Gibbs采样直接从SLAM完全后验分布产生样本,充分利用了信息矩阵包含的不确定信息,粒子分布均匀,且保持了多样性,缓解了粒子退化现象。实验结果表明所提算法的粒子集能够更好地描述真实后验分布,显著提高了SLAM算法的估计精度。     

基于稀疏扩展信息滤波和粒子滤波的SLAM算法

针对传统粒子滤波算法单次迭代过程中仅应用到当前的信息,且小权值粒子代表的信息在重采样中被删除而导致信息不能充分利用的问题,提出了稀疏扩展信息滤波和粒子滤波相结合的同时定位与地图创建（SLAM）算法,信息矩阵记忆了机器人位姿的历史信息,应用Gibbs采样重新获得粒子集,使粒子集能够更好地描述后验分布,提高算法的状态估计精度。大量的Monte-Carlo仿真实验验证了该算法中机器人定位精度较FastSLAM2.0算法提高80%左右。     

一种基于AMPF和FastSLAM的复合SLAM算法

为了改进快速同时定位和地图创建(FastSLAM)算法的粒子集性能、提高估计精度,提出基于AMPF和FastSLAM的复合SLAM算法.将辅助边缘粒子滤波器(AMPF)与FastSLAM架构相结合,用AMPF估计机器人位姿,单个粒子的位姿提议分布用无轨迹卡尔曼滤波估计.设计与AMPF和FastSLAM架构均兼容的采样方法和粒子数据结构,在FastSLAM框架下用扩展卡尔曼滤波递归估计地图.实验表明,该算法的粒子集性能比FastSLAM 2.0算法好,并且它的位姿估计精度高于FastSLAM 2.0算法.此外,粒子数较少时,该算法的估计精度较高,从而可适当减少粒子数目来提高算法的计算效率.     

一种改进Unscented粒子滤波及其应用研究

UPF(Unscented粒子滤波)克服了UKF(Unscented卡尔曼滤波)不适用于非高斯噪声系统的局限性和PF(粒子滤波)重要性密度函数难以选取的缺陷,但是,其仍易受到异常扰动的影响,且在计算迭代中协方差阵可能失去半正定性.针对此问题,提出抗差自适应Unscented粒子滤波算法.首先介绍抗差估计基本原理和自适应因子模型,在充分吸收抗差估计、自适应滤波和UPF算法优点的基础上,提出RAUPF算法,然后将RAUPF算法应用到单变量非静态状态增长模型中进行仿真计算,并与EKF和UPF算法进行比较,并进行复杂度分析.结果表明,提出的算法估计精度高于EKF和UPF算法的估计精度,估计值更加接近真值.     

An Improved FastSLAM Algorithm Based on Revised Genetic Resampling and SR-UPF

FastSLAM is a popular framework which uses a Rao-Blackwellized particle filter to solve the simultaneous localization and mapping problem (SLAM). However, in this framework there are two important potential limitations, the particle depletion problem and the linear approximations of the nonlinear functions. To overcome these two drawbacks, this paper proposes a new FastSLAM algorithm based on revised genetic resampling and square root unscented particle filter (SR-UPF). Double roulette wheels as the selection operator, and fast Metropolis-Hastings (MH) as the mutation operator and traditional crossover are combined to form a new resampling method. Amending the particle degeneracy and keeping the particle diversity are both taken into considerations in this method. As SR-UPF propagates the sigma points through the true nonlinearity, it decreases the linearization errors. By directly transferring the square root of the state covariance matrix, SR-UPF has better numerical stability. Both simulation and experimental results demonstrate that the proposed algorithm can improve the diversity of particles, and perform well on estimation accuracy and consistency.     

A SLAM based on auxiliary marginalised particle filter and differential evolution

FastSLAM is a framework for simultaneous localisation and mapping (SLAM) using a Rao-Blackwellised particle filter. In FastSLAM, particle filter is used for the robot pose (position and orientation) estimation, and parametric filter (i.e. EKF and UKF) is used for the feature location's estimation. However, in the long term, FastSLAM is an inconsistent algorithm. In this paper, a new approach to SLAM based on hybrid auxiliary marginalised particle filter and differential evolution (DE) is proposed. In the proposed algorithm, the robot pose is estimated based on auxiliary marginal particle filter that operates directly on the marginal distribution, and hence avoids performing importance sampling on a space of growing dimension. In addition, static map is considered as a set of parameters that are learned using DE. Compared to other algorithms, the proposed algorithm can improve consistency for longer time periods and also, improve the estimation accuracy. Simulations and experimental results indicate that the proposed algorithm is effective.     

Adaptive square-root transformed unscented FastSLAM with KLD-resampling

The FastSLAM relies on particles sampled from the proposal distribution of underlying Rao–Blackwellized particle filter, and its performance is significantly influenced by the quality and quantity of the particles. In this paper, a new improved FastSLAM is proposed based on transformed unscented Kalman filter (TUKF) and Kullback–Leibler distance (KLD) resampling method. In the proposed algorithm, a square-root extension of TUKF is used to calculate the proposal distribution and to generate credible particles. In addition, during the resampling process, the minimum required number of particles is determined adaptively by bounding the KLD error between the sample-based approximation and true posterior distribution of the robot state. Both numerical simulations and real-world dataset experiments are used to evaluate the performance of the proposed algorithm. The results indicate that the proposed algorithm achieves higher estimation accuracy and computational efficiency than conventional approaches.     

Effective cubature FastSLAM: SLAM with Rao-Blackwellized particle filter and cubature rule for Gaussian weighted integral

Simultaneous localization and mapping (SLAM) is a key technology for mobile robot autonomous navigation in unknown environments. While FastSLAM algorithm is a popular solution to the large-scale SLAM problem, it suffers from two major drawbacks: one is particle set degeneracy due to lack of measurements in proposal distribution of particle filter; the other is errors accumulation caused by inaccurate linearization of the nonlinear robot motion model and the environment measurement model. To overcome the problems, a new Jacobian-free cubature FastSLAM (CFastSLAM) algorithm is proposed in this paper. The main contribution of the algorithm lies in the utilization of third-degree cubature rule, which calculates the nonlinear transition density of Gaussian prior more accurately, to design an optimal proposal distribution of the particle filter and to estimate the Gaussian densities of the feature landmarks. On the basis of Rao-Blackwellized particle filter, the proposed algorithm is comprised by two main parts: in the first part, a cubature particle filter (CPF) is derived to localize the robot; in the second part, a set of cubature Kalman filters is used to estimate environment landmarks. The performance of the proposed algorithm is investigated and compared with that of FastSLAM2.0 and UFastSLAM in simulations and experiments. Results verify that the CFastSLAM improves the SLAM performance.     

快速SLAM算法的一种新的滤波架构

FastSLAM算法是同时定位与地图创建领域的一类重要方法,UPF-IEKFFastSLAM2.0算法采用UPF估计机器人的路径,地图估计则采用IEKF算法。UPF算法使粒子向后验概率高的区域运动,提高了估计精度,并且UPF算法比普通粒子滤波算法需要更少的粒子数,因而可以降低计算复杂度;IEKF算法通过迭代观测更新过程来提高估计精度。仿真实验表明,当迭代次数小于等于2时,UPF-IEKFFastSLAM2.0算法的地图估计累计时间比UPF-UKFFastSLAM2.0算法短;当迭代次数为2时,其估计精度高于UPF-UKFFastSLAM2.0算法。综合考虑估计精度和计算复杂度,认为"UPF-IEKF"是一种更合理的FastSLAM算法滤波架构。