A deep reinforcement learning technique for bug detection in video games

The objective of this research is to design the deep reinforcement neural learning-based model that detects the bugs in a game environment. The model automates the bug detection and minimizes human intervention. It makes effective use of the Deep-Q-Network to design and develop the model ‘RLBGameTester’ for measuring the high dimensional sensory inputs. The model modifies the environment to intercept the game screen. It also adds faults to the game before submitting it to the Deep-Q-Network. It calculates the values of the loss function at different iterations. The differences in the values of the loss functions in a bug-free and the bug containing game environment point out the presence of a bug. It also locates the position where the bug appears. The proposed model is useful for multiple game environments with minimum customization. Its applicability for blurred as well as non-blurred inputs at different platforms proves its efficacy. Employing this model may prove a game changer in the game designing industry. © 2022, The Author(s), under exclusive licence to Bharati Vidyapeeth's Institute of Computer Applications and Management.