Legged robots

ROSPug

A ROS quadruped robot dog on NVIDIA Jetson Nano, with 12 servos, lidar SLAM and Gazebo simulation.

NVIDIA Jetson
Suggested for university

ROSPug is a quadruped robot dog built on ROS. Twelve 30kg high-voltage serial bus servos drive the shoulder, hip and elbow joints, and a dynamic balancing kinematics algorithm lets it switch between gaits. A Jetson Nano works with an MCU controller and is joined by a TOF lidar, an HD wide-angle camera, an IMU and an OLED display, for SLAM mapping and navigation, path planning, dynamic obstacle avoidance and climbing stairs. ROSPug supports Gazebo simulation for checking quadruped kinematics and path planning, and its Python code is open source and annotated.

Zeta42 is an authorised Hiwonder partner in the Middle East.

What it does

Built for.

12 joints in aluminium
Twelve servos across the elbow, shoulder and hip of each leg mimic a real quadruped. The body is aluminium alloy, the calf joints have metal bearings, and a link structure extends the legs' range of rotation.
Lidar SLAM
The lidar supports SLAM mapping, multi-point navigation, path planning and dynamic obstacle avoidance.
Vision with OpenCV
The camera tracks colours, KCF targets and faces, reads tags and follows red lines. ROSPug spots stairs to climb them, kicks a ball with PID tracking, and uses vision and lidar together to circle a post.
Two controllers
The Jetson Nano handles AI processing while an MCU takes care of high-frequency control.
Gazebo simulation
Built on ROS with Gazebo support, ROSPug lets students validate kinematics and path planning in simulation, with ROS-based inverse kinematics functions and parameter debugging software.
Control
Drive it from the WonderROS app on Android or iOS with a live camera view, from the graphical PC software or with a wireless controller.

Where it is used