Video: Figure 01 Robot Trains for First BMW Assembly Job

About six months ago, Figure revealed that its shiny silver humanoid robot had secured its first role at a BMW manufacturing plant. It has been gearing up for this moment, and a new video highlights its progress as a versatile worker

Currently, numerous AI-driven humanoids are in development, featuring various hardware and learning strategies. Despite their differences, they all aim to perform any physical task a human can, but more cost-effectively and reliably.

Advancing Robot Capabilities and Economic Potential

The more tasks these robots can master, the more units companies can potentially sell, which might separate economic growth from population figures and unlock a labor pool limited only by available resources.

While it seems like a distant dream, the hardware is surprisingly advanced. Even though the AI models training these robots are still in early development, they are progressing quickly

Early Adoption of Basic Tasks in Customer Operations

While a truly versatile general-purpose humanoid is still a long way off, companies are eager to have them perform basic, practical tasks in real customer operations soon.

This is likely more about collecting data and teaching customers to work with these cutting-edge AI robots rather than generating significant revenue. Car manufacturing seems to be a prominent industry for early humanoid robot adoption.

Tesla operates its expansive car manufacturing facilities where it can train, test, and derive early benefits from its Optimus robot.

However, Figure Founder Brett Adcock suggests that relying solely on internal testing might not challenge the Tesla robot enough, tweeting, ‘Having an external customer who can terminate our contract if we fail will provide Figure with a competitive edge.’

Today, we can observe the progress of the Figure robot’s training in the fully autonomous ‘BMW Full Use Case‘ video below:

Highlights of the 01 Robot’s Progress

While not the most groundbreaking humanoid robot video we’ve seen, there are several positives to note here: The 01 robot is learning to grasp large, intricate shapes using its rubber-tipped fingers. It’s acquiring the ability to maneuver them without colliding into objects – a skill I wouldn’t necessarily trust my own children with at this stage.

Equally noteworthy is its achievement in placing a large plate on a jig, despite the plate obstructing the two locating pegs needed to fit through the holes.

“While the robot picks up the two smaller pieces one by one, which is less efficient than how humans typically handle them in pairs, it accurately identifies orientation errors, aligns them with the jig, and places them precisely.” Moreover, it self-corrects by tapping them into place with the back of its hand upon realizing their misalignment.

“However, as one witty YouTube commenter notes, it seems the robot is being compensated by the hour, not by the task. Nevertheless, Figure deserves credit for not accelerating the footage, indicating that these movements will likely become quicker and more fluid over time.”

Evolutionary Progress from Initial Demonstration

This may not appear as a significant leap from the company’s initial real-world task demonstration in February, shown above. Both could be categorized as tasks involving ‘picking up and placing down,’ making this update feel more like an evolution than a revolution.

However, the 01 robot is now handling much more complex and diverse shapes, requiring more intricate grips. It’s not just placing a simple box on a conveyor belt; it’s aligning intricate parts to fit precisely into jigs.

While the video concludes with ‘BMW Group, Plant Spartanburg,’ it seems to have been filmed in Figure’s own facility on a testing setup. Assuming this is the case, it remains uncertain when the 01 will actually start working on the BMW production line.

It will be intriguing to observe the performance of these robots in practical work environments. Initially, they will be slow, cumbersome, and constrained, but as their AI models evolve, both their movement and the speed of skill acquisition will increase.

However, these robots must clearly demonstrate an economic advantage over human workers to revolutionize industries, enter mass production, and drive substantial societal change. That’s when things will become truly compelling!

Read the original article on: New Atlas

Read more: Robotic Faces Covered with Living Human Skin