ZGDino
AI and edge computing

The next generation of animatronic dinosaursPowered by AI and edge computing

Zigong craftsmanship paired with real-time computer vision, ASR voice interaction and biomimetic actuation.

AI animatronic dinosaur head with camera and sensor hardware
Dual cameraMic arrayServo jaw

A looping prop, or an exhibit that answers

A traditional animatronic repeats the same motion on a timer. Visitors watch a cycle, take a photo, and move on, and the exhibit cannot respond to the person standing in front of it.

The smart series keeps the frame, skin and sculpture work, then adds an edge computer that sees, listens and replies. Overseas buyers get the academic realism of a museum piece together with interaction they can demonstrate on the floor.

Existing build
Steel frame, silicone skin, servo movement
+
Added layer
Raspberry Pi 5 edge computer running the interaction stack
=
Result
A dinosaur that sees the visitor, hears the question and answers it

Three systems, one reaction

Detection, conversation and jaw movement share one session, so the answer arrives with a look and a matching mouth.

Visitor standing in front of an AI animatronic dinosaur exhibit
Vision and tracking

It looks at the person who is talking

Dual cameras detect faces in the frame, then the head follows on horizontal and vertical axes.

It hears the question

A microphone array with ASR passes the request to an AI conversation layer, so follow-up questions keep the thread.

Its mouth keeps up

A dedicated servo channel drives the jaw from the speech output, so the mouth opens and closes with the answer instead of a fixed loop.

Vision and tracking

Dual USB and CSI cameras with face detection turn the head toward the visitor in front of it.

Voice conversation

A microphone array with speech recognition feeds an AI conversation layer, so questions get answers.

Biomimetic lip sync

Servo channels drive the jaw in time with speech output, keeping the mouth believable while it talks.

Edge control box with circuit boards and servo wiring

Hardware inside the build

The control stack sits in the body of the model, wired to the existing frame and skin. The same architecture is used for head mechanics, jaw sync and the camera and audio front end.

Main computer
Raspberry Pi 5 running the interaction stack at the edge
Servo control
PCA9685 driver board with a multi-channel SG90 and high-torque servo array for head axes and jaw
Audio
Omnidirectional microphone array and full-range speaker
Vision
Dual high-definition capture units on USB and CSI
Safety and compliance
Control box and flame-retardant material specifications follow UL and CE practice for overseas venues; certification scope is confirmed per project

One dashboard for the venue team

Configuration and monitoring run from the local web interface. The same interface is the base for later upgrades, so a venue can add sensors, motor channels or language packs without rebuilding the controller.

Managed on the venue network, no client install
How the DinoAi control system works
  • Remote device configuration and behaviour logic setup
  • Live camera preview from the management page
  • Audio files, voice packs and playback settings
  • Wi-Fi and network diagnostics
  • Sensor and motor expansion for later upgrades

Expansion ports and motor channels are reserved for second-stage customisation: behaviour sequences, additional language packs and multi-camera setups.

Axes and interfaces at a glance

Head yaw, head pitch and jaw run on separate channels, with travel set per model. Sensor inputs and spare motor channels are reserved for later scenes and triggers.

Read the integration whitepaper
AxisMotionAvailability
Head yawTurns left and right to follow the visitorStandard, travel set per model
Head tiltNods up and down toward the visitor's heightStandard, travel set per model
JawOpens and closes with the speech outputStandard, driven by a dedicated channel
Eyelids and neck leanExpression accents and body leanPer model, when the shell allows

Where the smart series works

Each venue gets its own character rules and language pack, set before the exhibit opens.

Museums and science centers

An interactive exhibit that answers questions about the species it is built from, next to the specimen label.

Theme parks and family entertainment centers

A walkway attraction that notices visitors, follows them with its head and starts a conversation.

Commercial events and touring shows

A compact interactive piece for mall activations and pop-up exhibitions that draws a crowd without a narrator.

Buyer questions on the smart series

What the AI layer adds, what runs on the controller, and what to confirm before an overseas order.

A life-size animatronic dinosaur with an edge computer inside the body. It combines dual-camera vision tracking, speech recognition with AI conversation, and servo-driven jaw movement, while keeping the frame, skin and sculpture work of a standard animatronic build.

A traditional model repeats a motion loop on a timer and cannot respond to the person watching it. The smart series detects the visitor, turns the head toward them, hears a question and answers it, with the jaw moving in time with the speech.

A Raspberry Pi 5 mounted inside the model runs the interaction stack. The exhibit does not need a separate development workstation running at the venue, and configuration is handled through the local web interface.

Dual USB and CSI cameras feed a face detection model that locates visitors in the frame. The head then follows the detected person on horizontal and vertical axes, so the model looks at whoever is speaking to it.

An omnidirectional microphone array captures speech, automatic speech recognition turns it into text, and an AI conversation layer produces the reply through the built-in speaker. Follow-up questions keep the same thread within a session.

Yes. A dedicated servo channel drives the jaw from the speech output, so the mouth opens and closes with the answer instead of running a fixed loop. A PCA9685 driver board coordinates the jaw and head axes.

Through a local web dashboard for device configuration, camera preview, audio and voice packs, behaviour logic and network diagnostics. Expansion ports and motor channels are reserved for later customisation such as new behaviour sequences or multi-camera setups.

The control box and flame-retardant material specifications follow UL and CE practice for export installations. The exact certification scope is confirmed per project, so ask for the documentation that applies to your market before ordering.

The unit runs its own interaction stack and is not supplied as a DMX-native fixture. For venues that cue exhibits from a show-control system, the interface and command set are reviewed per project, and the agreed option is quoted with the build.

Daily management happens on the venue network through the local dashboard. Monitoring or integration beyond that network is scoped per project rather than included by default.

Behaviour sequences, voice packs and audio playlists are configured in the dashboard. Custom motion programming for the head axes and jaw is produced with the animatronic team as part of the order.

The head and jaw run on multiple servo channels through the PCA9685 driver, and the exact axis count depends on the species and head size. The axis map is agreed at design stage.

Detection in the standard build is camera based. Additional triggers such as proximity sensors or floor buttons are specified per project, and the required inputs are confirmed before production.

Get the technical specification sheet

Tell us the species, size and venue. We will send the control architecture, servo layout and the options for your floor.