Collaborative robots (cobots) are designed to operate around humans in a factory environment and to execute tasks in a human-like fashion. Interestingly, assembly tasks that are simple for humans to perform such as gripping, grasping, turning and rotating items require highly sophisticated reinforcement learning (RL) protocols and tactile sensors for cobots to repeatedly successfully perform. Visual programming routines in the onboard software enable operators to program a robot via a “train by demonstration” graphical user interface. However, the technology behind these easily programmed and controlled robots is complex.
The post Using tactile-based reinforcement learning for insertion tasks appeared first on The Robot Report.
