全天时复杂背景下地面侦察机器人的视觉手势遥操作控制的理论与方法研究

This project aims to meet the urgent demand of ground reconnaissance robot for vision-based posture teleoperation control, and carries out the research around the key problems of vision-based posture teleoperation control at all-time and on complex background, starting with the key problems that need to be solved urgently in the current research situation. Firstly, the real-time robust detection method of vision-based hand detection and the optimal estimation method of hand pose are studied, which can adapt to all-time complex background, in order to realize hand detection and pose estimation under illumination change, color similarity, attachment occlusion and texture interference. Then, the stability of the position and posture of the end-effector of the manipulator under fast traction is studied, which enables the reconnaissance robot to achieve rapid response and stable tracking of hand at all-time and in complex background. It enables soldiers to control the reconnaissance robot remotely in a natural, intuitive and efficient way, and improves the efficiency of human-robot cooperation in military tasks. The theory and method of vision-based posture teleoperation control studied in this project has broad application prospects in the field of human-robot interaction of military robots. It is expected that this research will provide a theoretical and technical basis for the application of human-robot interaction teleoperation control theory in ground reconnaissance robots, and will further improve and enrich the theoretical system of human-robot interaction teleoperation control.

本项目针对地面侦察机器人对视觉手势遥操作控制的现实需求，从目前研究现状中存在的亟待解决的关键问题着手，对全天时复杂背景下地面侦察机器人的视觉手势遥操作控制的核心问题展开理论研究。首先研究可适应全天时复杂背景的视觉手势的实时鲁棒检测方法和手势姿态的最优估计方法，实现在光照变化、颜色相似、附着物遮挡和纹理干扰下的手势检测与位姿估计；然后研究多自由度机械臂末端位姿在快速牵引下的稳定性问题，使侦察系统在全天时复杂场景下实现对手势的快速响应与稳定跟踪，使士兵能够以一种自然、直观、高效的遥操作控制方式实现对侦察机器人的远程控制，提升人机协作执行军事任务的效率。本项目中研究的视觉手势遥操作控制的理论与方法在军事机器人人机交互操作领域有广泛的应用前景。预计本项目的研究将会为人机交互遥操作控制理论在地面侦察机器人上的应用提供理论与技术基础，将会进一步完善和丰富人机交互遥操作控制的理论体系。