A Study of Electric Powered Wheelchair with Operation Assist System which does not Prevent Human Operation if Unnecessary

这项研究的目的是使电动轮椅智能化，并为那些难以或不可能驾驶传统轮椅的人开发一种移动辅助设备。为了做到这一点，已经研究了用于人类操作电动轮椅的辅助方法。通过自主避障和人工操作相结合的功能，实现了电动轮椅车的安全驾驶和轻松操作。本研究的目的是构建不妨碍人的操作的操作辅助系统。在本研究中，神经网络产生避障功能。通过根据轮椅附近障碍物的情况和轮椅的运行状态来改变神经网络的连接权值的方法，改进了神经网络的避障功能。首先，神经网络是通过使用数字计算机模拟器来进化的。第二，实验，使用protot ...更多信息 通过对神经网络模型的仿真研究，证明了变权值神经网络具有更好的避障能力，并在智能轮椅样机上安装了18个PSD传感器，实现了对障碍物的检测。为了弥补PSD传感器的弱点，提出了虚拟传感器的概念，它保留了以往障碍物的观测数据，虚拟地构建了多个合适的传感系统，提高了避障功能。最后，我们提出了新的方法称为辅助率，这是用来控制不必要的自主避障。基于与障碍物和轮椅的行驶状态有关的信息来计算表示操作辅助功能的必要性的辅助率。虚拟传感器系统和辅助率已在原型上实现，所提出的方法的有效性已被证明。少

The purpose of this research is to make electric-powered wheelchair intelligent and to develop a mobility aid for people, who find it difficult or impossible to drive a conventional wheelchair. To do this, an assist method for human's operation of electric-powered wheelchairs has been investigated. Safe drive and easy operation of electric-powered wheelchairs has been realized by combining a function of autonomous obstacle avoidance and human operation. The goal of our research is to construct the operation assist system, which does not prevent human operation if unnecessary.In this study, a neural network produces an obstacle avoidance function. By using an approach that connection weights of the neural network vary according to the condition of obstacles in the neighborhood of the wheelchair and the running state of wheelchair, we have improved the obstacle avoidance function. First, neural networks have evolved by using digital computer simulator. Secondly, experiments, using protot … More ypes implemented neural networks whose connection weights were given by numerical studies, have demonstrated that the neural network with variable connection weights exhibits the more excellent ability of obstacle avoidance.To detect obstacles, 18 PSD sensors are mounted on our prototype of intelligent wheelchair. In order to compensate weak point of the PSD sensor, we have proposed a concept of the virtual sensor, which preserves previous observation data of obstacles and constructs virtually several suitable sensing systems and improves obstacle avoidance function. Finally, we have proposed new method called the assist rate, which is used in order to control unnecessary autonomous obstacle avoidance. The assist rate, which indicates necessity of the operation assist function, is calculated based on information concerned with obstacles and the running state of wheelchair. The virtual sensor system and the assist rate have been implemented on a prototype and the effectiveness of the proposed method has been demonstrated. Less