This paper presents a fast global collision detection method for robotic machining of large complex components, aiming to quickly determine whether there is a collision between the robot and the surrounding environment during the whole machining process. Geometric analysis shows that there are always some trajectory points on the motion path of the manipulator that are more likely to collide than the surrounding points during machining. These trajectory points with the highest collision probability within a certain range are defined as the feature points of global collision detection, and are used to replace all trajectory points to perform global collision detection, thus greatly improving the efficiency of related operations while ensuring accuracy. Compare to the traditional discrete collision detection method with computational complexity O( $\text{n}^{2}$ ), the computational complexity of the proposed method is only O(n). Numerical analysis and application experiments verify the effectiveness of the proposed method. Note to Practitioners—Motion planning in robotic machining of large complex components usually needs to perform a lot of global collision detection. Existing methods generally have the problems of large calculation and low efficiency, which seriously affects the efficiency of motion planning. This is mainly because a single global collision detection usually includes no less than $n$ times of static collision detection, where $n$ is the number of trajectory points. In order to solve this problem, we present a new global collision detection method based on feature-point-set. It does not need to traverse all trajectory points for static collision detection, but only needs to detect a few feature points, that is, the trajectory points most likely to collide within a certain range. On the premise of ensuring the collision detection accuracy, the proposed method greatly reduces the execution times of static collision detection, and significantly improves the computational efficiency of global collision detection. Numerical analysis and experiments show that this method effectively improves the efficiency of motion planning in robotic machining of large complex components.

中文翻译：

---

基于特征点集的大型复杂零件机器人加工快速全局碰撞检测方法


本文提出了一种用于大型复杂零部件机器人加工的快速全局碰撞检测方法，旨在快速判断机器人在整个加工过程中是否与周围环境发生碰撞。几何分析表明，加工过程中机械手运动路径上总有一些轨迹点比周围点更容易发生碰撞。将这些在一定范围内碰撞概率最高的轨迹点定义为全局碰撞检测的特征点，并用其替换所有轨迹点进行全局碰撞检测，从而在保证精度的同时，大大提高了相关操作的效率。与计算复杂度为O($\text{n}^{2}$)的传统离散碰撞检测方法相比，该方法的计算复杂度仅为O(n)。数值分析和应用实验验证了该方法的有效性。从业者注意事项——大型复杂部件的机器人加工中的运动规划通常需要执行大量的全局碰撞检测。现有方法普遍存在计算量大、效率低的问题，严重影响运动规划的效率。这主要是因为单次全局碰撞检测通常包含不少于$n$次的静态碰撞检测，其中$n$为轨迹点数。为了解决这个问题，我们提出了一种新的基于特征点集的全局碰撞检测方法。 它不需要遍历所有轨迹点进行静态碰撞检测，而只需要检测少数特征点，即在一定范围内最有可能发生碰撞的轨迹点。该方法在保证碰撞检测精度的前提下，大大减少了静态碰撞检测的执行次数，显着提高了全局碰撞检测的计算效率。数值分析和实验表明，该方法有效提高了机器人大型复杂零部件加工运动规划的效率。