基于全方位视觉跟踪的机器人轮椅的研究

The objective of the proposed research is to develop a new Robotic Wheelchair System which makes use of the Ubiquitous Visual Sensor Systems (UVSS), currently under rapid deployment, and the maturing Internet and wireless communication technologies. The proposed innovative RWS departs completely from.the conventional approaches of autonomous robots which rely on on-board sensors to map locally their environment for navigating progressively and autonomously. Instead, the proposed RWS makes use of an innovative global monitor and control system empowered by UVSS, Internet, and wireless communications in combination with the on-board wheelchair sensors and control..In the proposed RWS the robotic wheelchairs will be controlled anywhere within the operational environment by both the global wheelchair monitoring system and by the wheelchair on-board autonomous systems. The purpose of the on-board autonomous systems is to detect the obstacles in the neighborhood of the wheelchair and to enable its autonomous navigation when the global control from RWS is not available due to delays and packets dropouts in the communication channel.????.To bring the proposed new robotic wheelchair system into reality, the proposed project will address key technical issues with the specific objectives as described below:.o Determine the optimal distribution of the video cameras using a geometric method to estimate the minimum number of cameras required to cover continuously the entire operational environment. Using the images from the multiple-video cameras, generate in real-time a global map of the entire operational environment for the purpose of continuously monitoring, navigating and controlling all robotic wheelchairs..o Integrate the mean-shift object detection and tracking algorithm with particle filtering for the real-time detection and tracking of the robotic wheelchairs. The objective of this innovative integration is to enable remotely continuous monitoring, navigation and control of the robotic wheelchairs for safe, robust and reliable operations in dynamically varying and complex environments..o Design and implement identification markers with special patterns and distinct features for wheelchair identification. The objective of these specially designed markers is to facilitate fast and robust target identification inside dynamically varying and complex environments surrounded by complicated backgrounds..o Investigate and implement a networked control architecture and feedback system for the remote control of multiple robotic wheelchairs without using dedicated links to each wheelchair. Integrate the local sensor and decision-making algorithms with the global navigation and networked control for the purpose of minimizing the effects of delays or packet dropouts in the wireless communication channel.

本项目研究是开发全方位的视觉跟踪系统，并通过全局视觉反馈控制来提升机器人轮椅的可靠，安全运行.研究内容包括：（1）多摄像头的最优化布局。就是用最少数量的视频摄像头覆盖整个工作环境。（2）全方位视觉环境下的机器人轮椅实时跟踪。根据机器人轮椅上标示的颜色或几何特征，实时跟踪机器人的位置和姿态。（3）全方位视觉环境下的机器人轮椅视觉反馈控制。在实时跟踪的基础上，实现机器人位置和姿态的反馈控制。控制的目的是要使机器人轮椅运动在全局规划的最优路径上。（4）远程监视和遥感控制。医护人员或轮椅使用者的监护人可以远程监视轮椅的运行状态，并能实现远程遥感控制操作。本项目利用现有的成熟技术和设备如视频跟踪，网络通讯与控制等，这样使得本项目实施的可行性更高。

本项目利用全方位视觉系统实现全局监视，导航，和机器人轮椅控制。全方位视觉系统跟踪和监视机器人轮椅的运行，网络服务器进行运算和传输控制信号。在项目中主要研究内容包括：①多个视频摄像头和网络服务器的布局。②视觉跟踪机器人轮椅。③视觉反馈控制。④远程遥感操作。重要结果包括：①完成机器人轮椅实验样机。电动轮椅系统:电动轮椅由两节12V铅酸蓄电池供电;采用双轮差速的驱动方式,用以操纵轮椅前进与转向。环境感知系统:智能轮椅配备有多种传感器构成的环境感知系统，包括由红外实现的障碍物检测，以及台阶边缘或深坑的防跌检测。控制系统：轮椅的控制系统以TI公司的2407DSP控制器和机载计算机组成，用来采集测距系 统、防碰系统与防跌系统的数据。②建立实时跟踪的数学模型，设计跟踪算法，实现在多摄像头下机器人轮椅的跟踪。③实现远程遥感操作。科学意义包括：①机器人轮椅使用更便捷，安全，可靠。由于采用全方位视觉系统，视觉摄像头可以根据机器人轮椅的颜色和形状特征跟踪其在运行环境的位置和姿态，从而使得全局服务器根据当前机器人轮椅的位置和姿态以及当前环境的动态信息实时规划出机器人轮椅的有效，安全，和可靠的路径。②可实现远程监视和遥控。由于采用全方位视觉系统，医护人员或轮椅使用者的监护人能通过全局网络和摄像头远程监视机器人轮椅的工作情况，如在有必要的时候，可以实现远程遥控机器人轮椅的运行。③轮椅成本降低。由于采用全方位视觉系统，全局网络服务器会对机器人轮椅给出全局的路径规划，而机器人轮椅本身不需要做出局部的路径规划，这样就减少了传感器在机器人轮椅的安装，从而降低了机器人轮椅的成本。④多项技术的融合。本项目涉及到多项技术的融合，其中包括多视觉传感器区域覆盖技术，视觉跟踪技术，具有延迟的视觉反馈控制技术，和网络控制技术等。