Disney Enterprises received 1 granted patent this quarter in the AI & Machine Learning category.
The patent covers technology for training virtual character motion controllers using signals from game controllers, addressing the disconnect between training data and real-time player input. This approach enables smoother and more natural character animation by incorporating actual controller input into the AI training process.
Disney's single AI and machine learning patent tackles a persistent challenge in video game animation: the mismatch between how AI models are trained and how they operate during actual gameplay. The system uses player input signals from game controllers directly during the training process, mapping these control inputs to pre-recorded motion sequences so the neural network learns on data that resembles what players will actually provide in real time. This method sidesteps the domain shift problem that typically occurs when models trained on motion capture data encounter the different statistical patterns of live controller input. By aligning the training and inference data distributions from the start, the approach eliminates the need for additional domain adaptation layers or extensive relabeling of existing motion datasets.
All data sourced from USPTO patent filings. Google Patents may take several weeks to index recent publications. If a link is unavailable, search for the patent number at USPTO Patent Public Search.