Two Disney patent applications published on September 24 describe artificial intelligence systems for theme parks. One would watch animal habitats and alert guests to animal activity, including sightings of a favorite animal. The other would listen to ride equipment and flag sounds that differ from normal operation. A published application does not mean Disney plans to build either system.
Alerts When The Animals Are Out
The first application, Techniques for Automatically Monitoring Animal Behaviors and Generating Notifications, was filed on March 19, 2025. It describes cameras in an animal habitat, as well as on tour vehicles, boats and drones, feeding images to an animal monitoring platform. Machine learning models would identify the animals in view and what they are doing, and the platform would send notifications in real time.
Guests could receive those notifications as text messages or push notifications on phones and tablets, on screens or loudspeakers inside vehicles, or through wearable RFID devices such as wristbands. A guest’s user profile could list a favorite animal, and some notifications could go to all guests. Before sending anything, the system could score each sighting on factors such as how rare the viewing opportunity is, how close the guest is and whether the animal or behavior matches the guest’s interests. After a notification, a tour vehicle or boat could be driven toward the viewing area.
The filing also describes alerts for employees. Staff could be notified when an animal behaves unusually in a way that could indicate distress, acts aggressively, or blocks a vehicle path, which could halt a tour. One object detection model would draw boxes around the animals in an image and label each by species, and keypoints would mark body parts such as the head, shoulders and paws. Anomaly detection models could learn normal behavior for a species, an individual animal or a specific location and flag departures from it. The application also describes fine-tuning a multimodal model on images and text about specific animals, so it could answer a question such as “What animals are in the image?”
Listening For Ride Equipment Problems
The second application, Techniques for Detecting Anomalies in Audio Recordings for Predictive Maintenance, was filed on March 21, 2025. It covers amusement park attractions, including rides such as roller coasters, trains and trams, and names equipment such as a track switch, an actuator or a motor. A microphone with a cover made of a cylinder and fuzzy material would be pointed at the part. When a sensor signals that the equipment is in service, the system would record it. Each recording of a track switch moving could last five to eight seconds.
Two machine learning models would handle the audio. The first would be trained on voice recordings layered over equipment noise, so it could separate human voices, including screaming, talking and laughing, from the sound of the machine. The second, an autoencoder, would try to rebuild each new recording of the equipment. The system would compare the reconstruction errors from a period such as one day against a baseline from a period when the equipment was operating normally, using a Wasserstein distance metric to decide whether the change is an anomaly. An alert could include a graph of the reconstruction error for maintenance crews.
In the filing’s example, a track switch moves for eight seconds. The relative reconstruction error over the last four seconds of that movement is significantly above the baseline.
Another Disney application, published in August, would project a second show onto fireworks smoke.
For the latest news and updates make sure to follow us on social media! Facebook | Twitter | Threads | Instagram | TikTok
