Kangaroo is a bipedal research platform for exploring advanced control methods in legged locomotion, and in recent months the team at PAL Robotics has demonstrated the robot free-walking! The prototype platform, first created in 2021 continues to evolve thanks to the dedication of the Legged Robotics team. Read on to find out more about the latest developments and future plans for Kangaroo, and to see the latest demo videos.

Kangaroo design: lightweight and powerful for running and jumping

Kangaroo has been designed to be both lightweight and powerful, enabling the robot to perform agile manoeuvres like running, jumping, and withstanding impacts. In order to achieve these motions, during the design, we analysed several existing bipedal platforms and developed a novel leg design based on ball screw linear actuators with closed kinematic chains, which significantly improves hardware performance.

However, we faced a challenge as there are only a few control algorithms and libraries that support closed kinematic chains. To overcome this challenge, we implemented different approaches:

  • Defining the full model of the robot and implementing a library that computes all the transformations using virtual constraints.
  • Defining a simple model and integrating all the non-linear transformations in the ROS transmissions.

Free-walking, walking and balancing using Whole Body Inverse Dynamics Controller and a ZMP-based position control algorithm

Using those approaches we demonstrated Kangaroo doing the following: 

  • Free-walking using a ZMP-based (Zero moment point) position control algorithm
  • Walking using a ZMP-based (Zero moment point) position control algorithm
  • Balancing using the WBID (Whole Body Inverse Dynamics Controller) algorithm

We also recently took the opportunity to record these live demonstrations with Kangaroo. These video demonstrations of Kangaroo were possible thanks to the latest developments in software and hardware on the Kangaroo platform.  

Kangaroo Free-walking Demo

 

 

Kangaroo Walking Demo

 

 

Kangaroo Balance Demo

 

 

Previously we demonstrated Kangaroo jumping using WBID (Whole Body Inverse Dynamics Controller), in addition to other features previously only tested in simulation.

Kangaroo: ZMP walking algorithm 

Using this algorithm, we are able to solve a specific type of challenge related to walking by adjusting where Kangaroo’s feet are placed and the trajectory, allowing Kangaroo to walk smoothly.

In more detail – the Zero Moment Point (ZMP)-based Walking Pattern Generator (WPG) approximates Kangaroo as a non-minimum phase system known as Linear Inverted Pendulum Model (LIPM). The LIPM is stabilised using a Linear Quadratic Regulator (LQR) scheme in order to solve a linear problem over a trajectory, whose inputs are foot placements and ZMP trajectory and the output is a stable trajectory for the Center of Mass (CoM) to make Kangaroo walk. 

Kangaroo: use of force control 

Force control involves using sensors to measure forces and torques applied to the robot, in this case, so that Kangaroo is able to adapt to external forces. This is useful for collaborative robotics and tasks that require sensitivity to external forces.

In more detail – force control with Kangaroo is used to handle the physical interaction between the robot and the environment and ensure safe and dependable operation in particular around humans. The control goals include keeping the interaction forces limited to guarantee a particular force along the directions where interaction occurs.

Kangaroo: future plans and ROS 2

Our ultimate goal with Kangaroo is to achieve high dynamic manoeuvres using RTO (Robot Trajectory Optimization) and MPC (Model Predictive Control). We believe that Kangaroo’s advanced capabilities will pave the way for future developments in the field of bipedal robotics, and we are excited to continue our research and innovation in this area.

ROS 2

The team at PAL Robotics will soon be moving to ROS 2 to get the latest benefits of ROS 2 Control with Kangaroo. ROS 2 is highly configurable, meaning developers can tailor it to their robot’s unique characteristics and the environment. One of the key advantages of the new ROS 2 is that it is more robust, flexible, and scalable. It addresses some of the key challenges that have arisen as robotics technology has advanced in recent years. Watch this space for more news.

Public simulation 

We are also working on public simulation for Kangaroo, so that you can try the platform for yourself – this will be available soon. In the meantime, you can view Kangaroo on ROS Wiki here.

Meet the Kangaroo team at Humanoids 2023 in Austin, Texas!

Since 2000 the IEEE-RAS International Conference on Humanoid Robots (Humanoids) has brought together robotics experts from around the world to discuss the latest developments in the world of humanoid robots. This year the conference will be heading to Austin, USA  – from 12-14 December – and the PAL Robotics’ team are joining as gold sponsors. We are also taking part in a number of workshops at the event – you can find out the latest news on these by following us on LinkedIn. We encourage you to come and meet the team at our booth!

If you would like to find out more about PAL Robotics and our humanoid biped research platforms including the progress of Kangaroo, visit our blog to stay up-to-date, and to ask any questions, don’t hesitate to contact us.