BEST Robotics – Laika, the quadruped robot with a tensegrity spine

Featured image of 2017 capstone team members, mentors and NASA scientists: Alice Agogino (faculty mentor), Drew Sabelhaus (doctoral student mentor), Lara Janse van Vuuren (MEng student), Huajing (Shirley) Zhao (MEng student), Asher Saghian(MEng student), Shu Jun Tan(MEng student), Robel Tweeldebirhan (MEng student), Jonathan Bruce (NASA Ames roboticist) and Adrian K. Agogino (NASA Ames roboticist). This team won the 2017 Fung Institute Mission Award.

Laika, the quadruped robot with a tensegrity spine

Master of Engineering students for 2017-2018: see bottom of this page for information about opportunities for this project.

The Berkeley Emergent Space Tensegritites Lab is researching flexible tensegrity spines for walking robots. Robots with flexible spines have many potential advantages over those with rigid body structures. Motion between a robot’s hips and shoulders could allow for more complex and efficient locomotion for quadrupeds, as well as greater ability to traverse unknown terrain and interact with unknown environments while keeping stable and safe. This project, the walking quadruped Laika, is designed to utilize a tensegrity spine as its backbone.

We’ve worked on Laika’s spine itself, the Underactuated Lightweight Tensegrity Robotic Assistive Spine (ULTRA Spine), as well as the supporting hardware for Laika’s legs.

A test that shows Laika’s spine in hardware is shown below. With a stiff spine, when the robot encounters an obstacle underneath its foot, it tips over.

However, when the spine actuates, and rotates, it can balance itself on this uneven terrain:

We’ve simulated this type of spine rotation in our NASA Tensegrity Robotics Toolkit simulator:

During the 2016-2017 year, students worked on designs for Laika’s legs (still under construction:)

Leg design 1 for Laika, based on the Cheetah Cub from EPFL.
Leg design 2 for Laika, based on the Ghost Robotics Minotaur.

 

 

 

 

 

 

 

 

 

 

The 2016-2017 Master of Engineering team also worked on new topologies for the spine, seen in this video:

 

During this past year, we’ve also been working on control systems for the spine using model-predictive control (MPC). The following video shows a bending motion for the spine in simulation using MPC:

Contact Drew Sabelhaus for more information: apsabelhaus at berkeley dot edu.

ULTRA Spine Team Picture 2016-04-27

Many thanks to the 2015-2016 ULTRA Spine team! From left to right: Jorge Vizcayno, Ankita Joshi, Zeerek Ahmad, Drew Sabelhaus, Lara Janse van Vuuren, Abishek Akella, and Heeyeon Kwon.

Master of Engineering 2017-2018 students:

We’re looking for 3-4 students to work on the following projects for the spine for the upcoming year:

  1. Electronics for motor control of the spine. We’re hoping to implement some of our control results in hardware this year. This project will involve making a circuit to control the motors in the spine, and writing software on the embedded microcontroller that tracks motor movement according to the lengths of cables from the controller.
  2. Leg hardware prototypes. We’ve done background research on leg designs, and this year, we are hoping to prototype Laika’s legs, and have the robot take its first steps.
  3. Spine hardware prototypes. Our team has developed new geometries of the spine in simulation and with rapid-prototyped hardware. This year, we are hoping to prototype some of these in higher fidelity in hardware, test them in comparison with the simulation, and use them in combination with the robot’s legs.

Students in the Controls or Design subject areas will be eligible for these projects. If you are interested, come see us at the capstone info fair on Friday, August 18th, 2017.

Also be sure to check out the spherical tensegrity team for projects opportunities!