A  B  S  T  R  A  C  T：The complexity of multi-axis, milling and drilling CNC machines and the demand of high precision for complex parts production increase the importance of safe and efficient tool-paths generation during manufacturing. In these machines, multi-tool working in the same area, or static un-programed machine elements, such as fixtures, another work-part may cause collision problems. It is important to know the collision risks in advance during production in order to avoid unexpected production stops, and machinery damage. This research study is focusing on automatic collision detection and avoidance for safe non-functional (rapid) tool-path generation in a static 2.5D milling or drilling machining environment, as an initial step toward complex, dynamic multi-axis machine-tool manufacturing. A 3D vision based Time of Flight (ToF) sensor provides the depth information about the manufacturing scene that are exploited by the method presented for taking an effective decision to automatically detect and avoid collisions in order to achieve safe tool-path during production. The concept presented opens up new areas for research and application of ToF camera in a CNC manufacturing environment for tool-path planning. The results obtained are for traversal safe tool-paths in a static environment, which will be adapted to more complex and dynamic real machining scenarios by integrating it with STEP-NC technology in future.

Introduction：Multi-axis machines are known for high speed production, autonomy, flexibility and precision with less human intervention. The fast growing machine complexity [1], increases the problems of safe tool-path generation and collisions between  production tool and the work part, among different production tool or among moving machine components or any other static machine elements. These collisions may cause severe damage to the machine itself, its precision as well as unexpected production stops that may add additional cost to the manufacturing [2].

Collision detection and avoidance has been a problem for multi- axis machines since many years [3]. Many researchers have been focusing on functional trajectories [4–8], where a direct contact with the work-piece is already investigated. Despite high speeds and  un-predictable  risk,  non-functional  trajectories  [9–12],   less work has been done to investigate rapid displacement in multi- axis machining environment. This paper focuses on rapid displacement tool-paths (non-functional trajectories), where collisions risks and damage could be comparatively  higher.  A rapid decision is required for collision detection and avoidance in multi-axis environment in order to avoid any un-wanted situation during  machining.

Modern multi-axis machines are able to detect collisions but they leave the decision process of its avoidance to the human operators. When a collision is detected a rapid response system or approach should help these machines to take an efficient decision for its avoidance in order to achieve a non-stop production and higher autonomy. In the case of any obstacle element, not properly programed such as different tool length, foreign elements, material left on the work part, misplaced fixtures or any other machine element, wrong tool placement in the magazine, a system  is needed to detect the real environment and decides a safe path.

Collision detection and avoidance is an important problem for many industrial applications. Many researchers have been working for collision detection and avoidance problem in robotics. An elastic string and potential fields based algorithm presented by Lee [13] and Petiot [14] has produce good results in robotics,  whereas rapid decision for high speed tool-paths in multi-axis machine needs a more on-time and adapted approach. A generalized pattern approach [15] searches safe trajectory in a search space, which may take longer for rapid decision in multi-axis machines. An A* method extension of Edsger Dijkstra’s algorithm, mostly used in robotics [16], determines the cost optimal path from starting position to the target location is considered as an effective method but the disadvantage of this method lies in the local minima and also knowing the context of machine may not favor the use of full space  search algorithms. 铣削和钻削数控机床英文文献和中文翻译:http://www.chuibin.com/fanyi/lunwen_206129.html