摘要
估计能源动力机器人在不可预测环境中的碰撞安全距离对保障其安全性极为关键。然而,机器人运动距离的估计通常是个时间成本高的过程,且要求传感器必须具备高精度。为此,我们采用了一种结合高斯过程回归与正向运动学核方法的策略,此方法能够更有效且准确地估量碰撞距离。通过实验验证,此策略即便在噪声较多的复杂环境中训练,其性能相较于传统标准几何方法在距离估计方面的效率提高了近70倍,准确度也提升了13倍,证明了其显著的优势。
Calculating the collision distance of energy powered robots is crucial for evaluating their safety in unpredictable environments. The estimation of robot motion distance is a time-consuming operation, and the accuracy requirements of the sensors used to measure the distance are strict. Therefore, a Gaussian process regression and forward kinematics kernel method is proposed to effectively and accurately estimate collision distance. Experimental verification shows that even if Gaussian process regression is trained in a noisy and complex environment, the proposed method can still shorten the distance by 70 times and improve accuracy by 13 times compared to traditional standard geometric methods.
作者
赵甜
钱晶
曾云
ZHAO Tian;QIAN Jing;ZENG Yun(School of Metallurgy and Energy Engineering,Kunming University of Science and Technology,Kunming Yunnan 650093;Yunnan University Hydraulic Machinery Intelligent Testing Engineering Research Center,Kunming Yunnan 650093)
出处
《软件》
2023年第10期27-29,共3页
Software
基金
国家自然科学基金(51869007)。
关键词
高斯过程
距离计算
运动学
gaussian process
distance calculation
kinematics
