Make science fiction into reality

Cutting edge AI and ML happens on DogBot. 

This is the platform for researchers to develop and benchmark their locomotion, perception and proprioception algorithms.

*Register to be the first to know when we release DogBot into the wild.

190619_DogBot_render

A new breed of robot

Comprehensive quadruped platform for testing and deploying new Machine Learning techniques.

LIGHT & SAFE

Lowest-in-class weight: 800mm high and less than 20kg, thanks to the carbon fibre chassis.

AGILE

Power needs control: 34kW of high-torque brushless motors results in huge capacity for agile motion-planning.

SURVIVE THE STREETS

Navigate relevant environments: Get out of the lab and test at a human-scale. The curbs, stairs and rough terrain need your algorithms.

AI READY

Interact with your data: 20,000Hz local motor control and up to 4kHz communication over EtherCAT at 100Mbit/s. With USB and CANbus included.

RUN FOR LONGER

Keep testing all day: Low-weight means less impact breakage, and batteries discharge slower to keep you and the DogBot running – and walking – for longer.

ROS INTEGRATION

Get into the code: Open-source controller in C++ with Python gait algorithms and ROS wrapper.

Contribute to significant AI research without waiting years for grants

Start simulating now using the URDF in Gazebo/RViz via ROS or Dart, Bullet, MuJoCo and other physics engines. Then transfer to DogBot to test your results.
https://github.com/ReactRobotics/DogBotV4

Register for preorder
Simple Startup

Safe operation procedures means you can take DogBot off the leash in university class rooms.

Maximise Access

Comprehensive Gazebo simulations let multiple students test ROS before they deploy.

Sensory Overload

Easily attach more sensors to the frame and plug in to the control boards through USB or Serial.

Time to play

The BMC2 motor controllers come with comprehensive open-source software that let you control every aspect of the robot in at the level of detail you want.

Download firmware at GitHub: https://github.com/craftit/DogBotSoftware

Packages

Get started – pre-order DogBot today.

  • Robot: quadruped chassis with built in locomotion
  • Power Supply: Run the robot on treadmills without using up battery supply
  • Spare Batteries: So you can charge while you use
  • And you get:
  • Manual: Thoroughly documented
  • Example Files: Walking code that works already
  • Software:
  • ROS stack with Python API
  • Open-Source low-level control software with C++ API
  • Gazebo simulation and URDF
  • Optional Flight Case: Transport DogBot to remote locations
  • Optional Stand with hanging straps: Start by walking in the air!
  • Optional Linear Overhead Rail: suspend DogBat while you are teaching it
  • Configure an optional perception and compute package when your robot is ready.

Who's using DogBot?

The Robot Intelligence Lab at Imperial College London are the first recipients of DogBot.

Dr. Petar Kormushev, Director

, Robot Intelligence Lab, Imperial College London

Get students excited about your course

Students want to use the best hardware.

Increase your sign-up for your Masters programme by ensuring you have the latest in robotic hardware available for them to use. This is a quadruped designed so that every lab can have a cutting edge lightweight robot for robotics and AI research.

Visit our YouTube page to see what Masters students at Imperial College London have been doing with DogBot: https://youtu.be/SRXOYvzKmkI

Pre-Order your DogBot now!

We’re busy testing DogBot so you can be one of the first to use the production version.

tell me the preorder release date