Deep reinforcement learning-based safe interaction for industrial human-robot collaboration using intrinsic reward function | Litlas