Crikey! An Affordable Robot Dingo From Down Under

Quadruped robots are robotic systems designed to mimic the locomotion and behavior of animals with four legs. These robots are typically equipped with advanced sensors, actuators, and control systems that allow them to navigate various terrains and perform a wide range of tasks that could traditionally only be accomplished by humans. Inspired by the natural movement and agility of animals like dogs and cheetahs, quadruped robots offer unique advantages in the field of robotics.

One of the key benefits of quadruped robots is their exceptional mobility. With four legs, these robots can traverse challenging terrains, including rough terrain, uneven surfaces, stairs, and even outdoor environments. Unlike wheeled or tracked robots that may struggle with obstacles or get stuck in certain conditions, quadruped robots can adapt to the environment and maintain stability while moving. This makes them suitable for applications such as search and rescue missions in disaster-stricken areas, exploration of unknown territories, or assistance in hazardous environments.

But before a quadrupedal robot can take to the world and help with any of these tasks, researchers first need to outfit them with the proper sensors and design the algorithms that make them tick. This can be a bit of a problem for university students and many other researchers, however, because these robots tend to fall into one of two categories — industry-grade robots that cost many thousands to tens of thousands of dollars, or hobbyist-grade robots that are inexpensive, but too small to do most real-world jobs.

A pair of student engineers at Monash University saw this as an opportunity to build a quadrupedal robot that is somewhere in between existing options in terms of functionality and cost. So they set out to design and build a robot for education and research purposes that was affordable, robust, expandable, and aesthetically pleasing. After several months of design and testing, they accomplished that goal with a robot they call The Dingo. It may not be as large as Boston Dynamics’ Spot, but it is quite capable for research, and it only costs about $1,500.

The Dingo weighs about three kilograms and is 25 centimeters tall, with a length and width of 38 x 25 centimeters. It can be directly controlled with a PlayStation controller or computer keyboard, and it also is equipped with a Raspberry Pi single-board computer for running more advanced control algorithms. An Arduino Nano was included for handling extra peripherals like analog sensors. With a LiPo battery, The Dingo can continuously operate for about 25 minutes, and there is power to spare for additional components a user may need to add on for their specific use case.

It is possible to control the robot’s pitch, roll, yaw, and height using the included control system. The robot’s top speed is 30 centimeters per second, and it can move in any direction. And because The Dingo can carry a payload of up to 500 grams, it can physically carry lots of additional sensors, actuators, and more for expandability.

Each of the four legs is powered by a trio of servo motors, each capable of producing 35 kilogram-centimeters of torque. These motors are placed in a clever arrangement, in which one sits on top of the other, such that these heavy components can be tucked inside the robot’s body, which reduces leg inertia and makes for smoother movements. Full state feedback is provided in both joint and task space such that advanced control and decision making algorithms can be developed. And the motors are fan-cooled to increase their lifespan.

The robot body is in a nicely finished custom case that includes USB and Ethernet ports on the back, alongside a small LCD display that can be used for any number of purposes, like displaying the status of the battery. There is a button on top of The Dingo that immediately cuts off power to the actuators (but not the control systems) in case of trouble during experimentation.

Using the Gazebo simulation software, the team also created a virtual replica of their robot. This allows control algorithms to be tested in simulation, and also enables the collection of training data for reinforcement learning algorithms that can then be transferred to the physical hardware.

The Dingo is incredibly polished for a student project, and looks to be a very promising platform for research into quadrupedal robots where budget is a concern. The team hopes that their device will be widely used in the future, and towards that end, they have released their source code and CAD models.