Boston Dynamics bipedal robots complete parkour course

By Laura Berrill
Boston Dynamics showcase accompanied by behind-the-scenes explanation of robot development process

Boston Dynamics has published a pair of videos showcasing its Atlas robots completing a complex obstacle course that requires balancing on beams and vaults. The company's Atlas robotics program is a platform for its engineers to perform research and development on robotic sensory and perception systems. 

A sandbox environment for testing

In a sandbox environment - an isolated testing environment enabling users to run programmes or execute files without affecting the application or platform on which they run - Boston Dynamics tasked two of its Atlas robots with parkouring through various obstacles.

The routine entailed one of the two robots running up a series of banked plywood panels, broad jumping a gap, and running up and down a set of stairs. The second was programmed to leap onto a balance beam and follow the same steps as the first robot, but in reverse. Both robots completed the routine by performing synchronised backflips.

The parkour routine can be viewed on YouTube.

Adapting robotic behaviours

However, it was found that when implementing these behaviours, the Atlas robots crashed a lot, according to Boston Dynamics in a separate "inside the lab" video. The company added that when running the routine repeatedly, the robots are able to get the vault portion of the routine right about half of the time. On some of the runs, the robot loses its balance and falls backward after it jumps over the vault.    

Boston Dynamics said the routine itself was choreographed, but what made this routine different from previous iterations is that the robots had to adapt behaviours in their repertoire, based on what they saw.

"Atlas' moves are driven by perception now, and they weren't back then," Atlas team lead Scott Kuindersma explained in an accompanying blog post.

"For example, the previous floor routine and dance videos were about capturing our ability to create a variety of dynamic moves and chain them together into a routine that we could run over and over again. In that case, the robot's control system still has to make lots of critical adjustments on the fly to maintain balance and posture goals, but the robot was not sensing and reacting to its environment," he went on.

According to Boston Dynamics, this meant engineers could create a smaller number of template behaviours and did not need to pre-program jumping motions for all possible platforms.

Previously, the Atlas robots were essentially blind when performing demonstrations and could only succeed when going through obstacles in unchanged environments.

While the two robots completed the course, Boston Dynamics engineers said the design still has room for improvement as one of the robots stuttered when pumping its arm in celebration after the routine.

And when performing the pumping movement, it also stumbled a bit, the company said.

Two months ago, the robotics company was acquired by Hyundai. The South Korean conglomerate acquired an 80% controlling interest in Boston Dynamics for $1.1 billion.



Featured Articles

Toshiba Takes Another Step to Ushering in Embodied AI

Toshiba's Cambridge Research Lab has announced two breakthroughs in Embodied AI alongside a new group to renew focus on the tech

Why AWS is Investing $230m in Credits for Gen AI Startups

Amazon is investing US$230m in AWS cloud credits to entice Gen AI startups to get onboard with using its cloud services

How Retrieval Augmented Generation (RAG) Enhances Gen AI

RAG is a technique that promises to improve the way Gen AI fetches answers and provide business with a more reliable use case for client-facing uses

Synechron’s Prag Jaodekar on the UK's AI Regulation Journey

AI Strategy

LGBTQ+ in AI: Vivienne Ming and the Human Power of AI

Machine Learning

Samsung’s AI-Era Vision Coincides With its New Chip Tech

AI Strategy