机器人学研究室知识库

如何构建准确可靠的环境模型一直是移动机器人的研究热点。2.5维地图是环境建模常用的环境描述方法。但是在野外环境中,由于机器人运动过程中存在的振动以及传感器的误差,造成高程地图出现偏差,偏差的范围将直接影响机器人路径规划的成败。传统的概率估计方法易于实现高程值的估计,但无法获得高程的可靠范围。为了解决这一问题,本文提出了一种基于鲁棒高程边界的地面机器人环境建模方法。首先构建传感器测量模型并获取测量误差的边界描述,在此基础之上,利用集员估计方法实现高程误差范围边界的估计,并给出了可通过性分析,最后利用实验验证了所提方法的可行性与有效性。与传统的环境感知方法相比,本方法可以获取可靠的高程范围,提升环境模型的鲁棒性和可靠性。

Accurate and reliable environment model is a key issue for mobile robots. However，in the field environment, due to the vibrations and sensor errors, the elevation of the 2. 5D map is always deviated. Focused on this problem, a robust elevation bound based environment modeling method is proposed in this work. In the first，the laser measurement model is researched and built. Based on the model, the bound description of measurement error is analyzed for describing the area of the measurement error. In the second, the estimation of elevation bound is realized by the extended set-member filter，which shows the elevation bound of the 2.5D map built by the mobile robots. Then the traversability of mobile robots is analyzed based on the elevation bound. In the last，the feasibility and effectiveness of the proposed method are verified by experiments. Compared with the common EKF based environment modeling methods,the robustness and reliability of the environmental map can be improved due to the reliability elevation bound.