摘要
机器人与机器学习每日新闻-机器人与自动化的最新研究结果已经发表。根据NewsRx记者在加州斯坦福特的新闻报道，研究表明，“基于学习的机器人感知模型存在两个不同的错误来源：人工和认知。随机不确定性产生于固有的噪声训练数据，很容易从训练过程中的残余误差中量化。”这项研究的财政支持来自国防高级研究计划局(DARPA)。新闻记者从斯坦福大学的研究中获得了一句话：“相反，认知不确定性产生于训练数据的缺乏，适用于分布外的运行状态，并且难以量化。现有的状态估计方法大多通过线性噪声模型处理偶然不确定性，但忽略了认知不确定性。”我们提出：(i)认知卡尔曼滤波器(EpiKF)将认知不确定性纳入学习感知模型的S tate估计，(ii)认知信念spa ce planner(EpiBSP)，该planner(EpiBSP)建立在EpiKF的基础上，规划轨迹以避免高认知不确定性和偶然不确定性区域。“倒置”从测量中预测状态的已学习的每感觉模型。我们将这两个模型组合在一个抽样方案中，以给出一个校准良好的外感和随机不确定性组合在线估计。”
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – Current study results on Robotics - Ro botics and Automation have been published. According to news reporting from Stan ford, California, by NewsRx journalists, research stated, “Learningbased models for robot perception are known to suffer from two distinct sources of error: al eatoric and epistemic. Aleatoric uncertainty arises from inherently noisy traini ng data and is easily quantified from residual errors during training.” Financial support for this research came from Defense Advanced Research Projects Agency (DARPA). The news correspondents obtained a quote from the research from Stanford Univers ity, “Conversely, epistemic uncertainty arises from a lack of training data, app earing in out-of-distribution operating regimes, and is difficult to quantify. M ost existing state estimation methods handle aleatoric uncertainty through a lea rned noise model, but ignore epistemic uncertainty. In this work, we propose: (i ) an epistemic Kalman filter (EpiKF) to incorporate epistemic uncertainty into s tate estimation with learned perception models, and (ii) an epistemic belief spa ce planner (EpiBSP) that builds on the EpiKF to plan trajectories to avoid areas of high epistemic and aleatoric uncertainty. Our key insight is to train a gene rative model that predicts measurements from states, ‘inverting’ the learned per ception model that predicts states from measurements. We compose these two model s in a sampling scheme to give a well-calibrated online estimate of combined epi stemic and aleatoric uncertainty.”
NETL
NSTL