Enhanced online programming for industrial robots

摘要

The use of robots and automation levels in the industrial sector is expected to grow, and is driven by the on-going need for lower costs and enhanced productivity. The manufacturing industry continues to seek ways of realizing enhanced production, and the programming of articulated production robots has been identified as a major area for improvement. However, realizing this automation level increase requires capable programming and control technologies. Many industries employ offline-programming which operates within a manually controlled and specific work environment. This is especially true within the high-volume automotive industry, particularly in high-speed assembly and component handling. For small-batch manufacturing and small to medium-sized enterprises, online programming continues to play an important role, but the complexity of programming remains a major obstacle for automation using industrial robots. Scenarios that rely on manual data input based on real world obstructions require that entire production systems cease for significant time periods while data is being manipulated, leading to financial losses. The application of simulation tools generate discrete portions of the total robot trajectories, while requiring manual inputs to link paths associated with different activities. Human input is also required to correct inaccuracies and errors resulting from unknowns and falsehoods in the environment. This study developed a new supported online robot programming approach, which is implemented as a robot control program. By applying online and offline programming in addition to appropriate manual robot control techniques, disadvantages such as manual pre-processing times and production downtimes have been either reduced or completely eliminated. The industrial requirements were evaluated considering modern manufacturing aspects. A cell-based Voronoi generation algorithm within a probabilistic world model has been introduced, together with a trajectory planner and an appropriate human machine interface. The robot programs so achieved are comparable to manually programmed robot programs and the results for a Mitsubishi RV-2AJ five-axis industrial robot are presented. Automated workspace analysis techniques and trajectory smoothing are used to accomplish this. The new robot control program considers the working production environment as a single and complete workspace. Non-productive time is required, but unlike previously reported approaches, this is achieved automatically and in a timely manner. As such, the actual cell-learning time is minimal.