Yahboom DOGZILLA S1/S2 12DOF AI Robot Dog for Raspberry Pi 5 (ROS2 Humble) Vision + LiDAR (S2)

Overview

DOGZILLA S1/S2 is a desktop-level 12DOF AI robot dog platform for Raspberry Pi 5 (ROS2 Humble), designed for embodied intelligence learning and robotics development. It integrates AI visual interaction, robot kinematics (inverse kinematics), and sensor feedback to support omni-directional movement and six-dimensional attitude control. The system includes an all-metal aluminum alloy body, built-in IMU and joint position sensors for real-time feedback on posture, joint angles, and torques, plus a camera for AI vision applications.

DOGZILLA also introduces multimodal large model capabilities, supporting intelligent voice dialogue and scene understanding. For questions about configuration, selection, or technical support, contact https://rcdrone.top/ or email support@rcdrone.top.

Key Features

• 12DOF bionic quadruped motion with posture self-balancing adjustment.
• All-metal aluminum alloy body with shock-absorbing silicone foot bar and ABS wear-resistant foot end.
• Sensor feedback: 6-axis IMU attitude sensor and joint servo angle readback / joint position sensors.
• AI vision functions (ROS2 + OpenCV) such as label recognition, face detection, target tracking, and visual line patrol.
• Multiple control methods: APP, handle controller, web pages, computer keyboards; supports APP mapping navigation.
• Built-in 20 bionic action groups (examples shown): Handshake, Sit down, Looking for food, Stretch oneself, Mark.

AI Large Model & Embodied Intelligence

Three AI Large Models

• Large Language Model: Real-time connection; understands text instructions and responds flexibly (examples shown: Text generation, Q&A, Science summary).
• Voice large model: AI large model voice module supports real-time conversion between voice and text, semantic understanding, intelligent voice Q&A, and voice control.
• Visual large model: High-definition camera can identify and analyze image content; supports multi-modal applications of pictures and text, including generating pictures from voice commands.

Embodied Intelligence Functions (examples shown)

• Autonomous line tracking: Identifies and tracks lines of various colors in real time.
• Embodied intelligent robot dog: Understands user instructions and responds with different feedback combined with visual recognition content.
• Multimodal Large Model + SLAM Mapping Navigation: S2 version exclusive function (see LiDAR section).

LiDAR Functions (S2 Only)

• TOF laser LiDAR for SLAM mapping navigation; supports LiDAR following and avoiding.
• LiDAR module specified as MS200 laser LiDAR (also shown as MS200TOF laser lidar), with 360° fast environment scanning.
• Functions shown: Navigation obstacle avoidance, Cartographer mapping, Mobile APP mapping navigation, LiDAR patrol, LiDAR follow, LiDAR guard, LiDAR avoid.

Note: The TOF laser LiDAR functions are indicated as only available for S2.

AI Visual Recognition Functions (examples shown)

• Mediapipe development
• Tag recognition
• AR vision
• Tag tracking
• Color recognition
• Face detection
• Target tracking
• Obstacle recognition
• QR code recognition
• Visual tracking

Also stated: 10+ AI visual recognition technology solutions.

Specifications

DOGZILLA Series Comparison (as provided)

Raspberry Pi 5 (optional, as stated)

• Performance stated as 2~3 times that of Raspberry Pi 4B
• Computing power: About 500GFLOPS
• GPU: Broadcom Videocore VII
• CPU: 64-bit 2.4 GHz quad-core

Hardware Structure (labeled parts shown)

• MS200 laser lidar (Just for S2)
• HD camera
• AI large model voice module
• Speaker and speaker base
• All-metal aluminum alloy body
• Raspberry Pi 5 board (Optional)
• OLED display
• ESP32 high performance co-processor
• Serial bus servo
• Lithium battery pack
• Silicone foot rod; ABS wear-resistant foot end

Handling note shown: Servo is a vulnerable item, do not press it.

Software & Development

• ROS2 system (ROS2 Humble stated in product title), Python programming support, RVIZ simulation.
• ROS2 + OpenCV workflows for AI vision functions (examples listed above).

Applications

• Education and scientific research
• Artificial intelligence experiments
• Service robot prototyping and embodied intelligence exploration

Tutorials & Videos

Tutorial link:http://www.yahboom.net/study/DOGZILLA

Details

A desktop-ready 12DOF robot dog platform built for Raspberry Pi robotics learning, AI vision, and embodied intelligence projects.

Designed for ROS2 development with options like SLAM mapping navigation (S2) alongside AI vision and tutorial resources.

12 degrees of freedom plus sensor feedback help support stable gait control, posture adjustment, and kinematics experiments.

Ongoing hardware and software upgrades add features such as voice interaction, Raspberry Pi 5 support, and S2 LiDAR navigation.

Core capabilities include AI vision, bionic action groups, multiple control options, and ROS2 tools for development and learning.

Built-in camera and voice hardware enable multimodal interaction for tasks like visual recognition, voice dialogue, and feedback.

Multimodal large model features support text Q&A, voice control, and camera-based scene understanding workflows.

Embodied intelligence functions connect perception and motion, from tracking behaviors to mapping navigation on S2.

AI vision demos cover common OpenCV/ROS2 learning topics such as recognition, tracking, and obstacle awareness.

S2 adds a TOF laser LiDAR module for SLAM mapping, autonomous avoidance, and LiDAR following behaviors.

Use the series comparison to choose between S1 and S2 configurations, including LiDAR and mapping-related features.

A library of bionic action groups helps demonstrate gait control and interaction behaviors during practice and lessons.

All-metal construction pairs with IMU and joint angle feedback to support more repeatable posture and movement testing.

Optional Raspberry Pi 5 support provides stronger compute for ROS2 development and onboard AI workloads.

A modular internal layout makes it easier to understand key components like camera, battery pack, and optional S2 LiDAR.

ROS2 + RViz workflows support inverse kinematics analysis, gait planning, and simulation alongside real robot testing.

Control options range from mobile app and mapping navigation to web control, gamepad handling, and keyboard inputs.

Teaching mode allows one robot to act as a master so another can follow the same leg motion for quick demonstrations.

Yahboom DOGZILLA S1/S2 comes with access to systematic tutorial courses via the provided online link.

DOGZILLA S1/S2 includes a broad set of AI and control modes—such as ROS2/OpenCV vision, face tracking, line following, and voice control—for flexible projects.

DOGZILLA control tutorial folders cover basic and advanced topics like color recognition, face tracking, QR code control, and voice interaction.

Yahboom DOGZILLA S1/S2 comes with organized tutorials for LiDAR mapping navigation, voice control, and ROS2 basic videos with English subtitles.

DOGZILLA S1/S2 dimensions are provided for power-on and power-off positions, including an approx. 246–250 mm length and 93–170 mm height range for fit planning.

DOGZILLA S1/S2 uses Raspberry Pi 5 ROS master control with 12DOF bus servos, a 5MP camera, and a 7.4V 5000mAh battery.

The DOGZILLA S2 LiDAR module lists a 360° scanning angle with 4,500 points/s and a 0.03–12 m distance measurement range.

DOGZILLA S1/S2 kits include the main body parts, gamepad controller, 64G TF card with reader, speaker, camera module, and basic tools and screws, with the Raspberry Pi board listed as optional.

Optional aluminum box packaging provides a protective carrying case, and the DOGZILLA S2 version lists an MS200 laser lidar with serial adapter board and connecting cable.