基于多模态数据和粒子滤波器的移动机器人目标跟踪方法

A Mobile Robot Target Tracking Method Based on Multimodal Data and Particle Filter

摘要:

在复杂环境中,目标可能受到遮挡、光照变化、背景干扰等因素的影响,导致移动机器人目标跟踪精度和效率低。为保证移动机器人目标跟踪效果,提出了基于多模态数据和粒子滤波器的移动机器人目标跟踪方法。采集移动机器人目标多模态数据,通过畸变校正、去噪、增强等步骤,完成移动机器人目标多模态数据的预处理。从边缘、颜色、纹理等多个方面,提取移动机器人目标多模态数据特征。以提取特征为研究对象,通过粒子滤波器的训练,得出移动机器人跟踪目标的检测结果,最终通过实时目标位置的更新,实现移动机器人目标跟踪。实验结果表明,在单目标和多目标两种实验场景下,设计方法的跟踪误差和跟踪时延较低,能够有效提高移动机器人目标跟踪精度和效率,具有较好的移动机器人目标跟踪效果。

Abstract:

In complex environments, targets may be affected by factors such as occlusion, changes in lighting, and background interference, resulting in low tracking accuracy and efficiency of mobile robots. In order to ensure the effect of mobile robot target tracking, a mobile robot target tracking method based on multimodal data and Particle filter was proposed. Collect multimodal data of mobile robot targets, and complete the preprocessing of multimodal data of mobile robot targets through steps such as distortion correction, denoising, and enhancement. Extract multimodal data features of mobile robot targets from multiple aspects such as edges, colors, and textures. Taking the extracted features as the research object, the detection results of the mobile robot tracking target are obtained through the training of the Particle filter. Finally, the mobile robot target tracking is realized through the real-time target position update. The experimental results show that in both single target and multi target experimental scenarios, the design method has low tracking error and tracking delay, which can effectively improve the accuracy and efficiency of mobile robot target tracking, and has good mobile robot target tracking performance.