Robots have several applications in different fields of nowadays life as in sports training-assistance. One of these sports is Fencing that is an individual duel Olympic sport using a bladed weapon. A typical fencer’s training consists of practicing different techniques. This practice is achieved through three training approaches that are not best utilized due to some constraints referred to humans’ nature and capabilities. The aim of this paper is to develop a robot-arm to be used in fencing training to overcome these constraints. This paper introduces a system for modelling fencing robot-arms using a fast and low-cost approach. The system estimates the angles values needed to drive a six degrees-of-freedom robot-arm that mimics a human-arm performing determined fencing movements. A simple and inexpensive system (Kinect) is applied for capturing the motions of the task and an Artificial Neural Networks model is used for transforming the captured motions into robot-arm movements through an inverse kinematics study. The proposed system was verified and validated, while the caused irregularities were referred to the changes in the lengths of the captured arm-segments and to the random error resulting from the depth measurement by Kinect using a single sensing camera.

中文翻译：

---

击剑运动训练用机械臂的建模

机器人在当今的不同领域中都有多种应用，如运动训练辅助。击剑运动是其中一项运动，它是一项使用剑刃武器进行的双人对决奥林匹克运动。击剑手的典型训练包括练习不同的技巧。这种做法是通过三种培训方法来实现的，由于涉及人的本性和能力的某些限制，这些方法不能得到最好的利用。本文的目的是开发一种用于击剑训练的机械臂，以克服这些限制。本文介绍了一种使用快速且低成本的方法对击剑机器人手臂进行建模的系统。该系统估计驱动六自由度机械臂所需的角度值，该六自由度机械臂模仿执行确定的击剑运动的人手臂。一个简单且便宜的系统（Kinect）用于捕获任务的动作，而人工神经网络模型则用于通过逆向运动学研究将捕获的动作转换为机器人手臂的动作。所提出的系统得到了验证和确认，而所引起的不规则性是指被捕获的臂段长度的变化以及由Kinect使用单个传感相机进行深度测量而产生的随机误差。