Collaborative robots (cobots) offer increased inter-action capabilities at relatively low cost. However, in contrast to their industrial counterparts, they inevitably lack in precision. The latter can generally be improved but it comes at the expense of time-consuming calibrations that need to be performed regularly. In addition, modeling errors at the task level rapidly accumulate as tasks change and robots are re-programmed, adding to the robots’ own kinematic and dynamic errors. These aspects strongly limit the application of cobots in tasks that require high precision. We discuss this challenge in the context of the COB’HOOK project that considers the use of cobots for the watch-making industry. We present the use case of picking up and assembling watch parts and discuss a solution combining high-precision laser sensing and an optimal control formulation that is used to plan relative trajectories.
Source: researchgate.net from Conference: R:SS Workshop on "Benchmarking Tools for Evaluating Robotic Assembly of Small Parts"