学术论文

      机器人姿态解算算法研究

      Approach to Robot Attitude Algorithm

      摘要:
      在机器人惯性导航研究中,针对传统滤波方法在非线性系统模型下误差大的问题,提出了一种基于改进粒子滤波的机器人姿态解算方法.粒子滤波精度较高且不受系统模型非线性程度的影响,与扩展卡尔曼滤波算法相比在非线性系统应用中有巨大的优势.使用扩展卡尔曼滤波对系统状态进行预测,使粒子分布向高似然区移动.对粒子滤波算法的重采样过程进行了改进,提升了算法的效率.不同的地面环境下系统噪声有较大变化,将地面环境信息作为观测信息融合到系统中,对算法参数进行实时修正能够获得更高的精度.实验结果表明,应用此算法进行姿态解算精度较高,且性能优异.
      Abstract:
      An attitude algorithm method based on improved particle filtering is proposed in the study of robot inertial navigation.The method is aiming at the problem that there is large error when the traditional filters works in nonlinear systems. Compared with the traditional extended Kalman filter,particle filtering has a huge advantage that it is not affected by the the level of nonlinear system model and it has high accuracy in nonlinear system. Extended Kalman filter is used to update the state, so that the particles can move to the high likelihood area. A improved resampling method is proposed to improve the efficiency. The system noise will change when the robot run at different ground conditions. The parameter is modified according to the ground conditions to obtain higher precision.The results of experiments show that the attitude reached high precision and performed well.
      Author: ZHANG Xiao-jun XU Zi-han YANG Shi-peng
      作者单位: 河北工业大学 机械工程学院,天津,300130
      刊 名: 机械设计与制造 ISTICPKU
      年,卷(期): 2018, (6)
      分类号: TH16 TP242
      在线出版日期: 2018年6月25日
      基金项目: 国家高技术研究发展计划(863计划)