Embodied AI robots for the future workforce

Could this AI framework transform the capabilities of tomorrow's machines?

Spotted: One of the central promises of AI is that it will someday be able to relieve humanity from strenuous and tedious tasks. Today, startup Swiss-Mile is drawing closer to that goal by connecting AI with the physical world to create truly autonomous robots.

Swiss-Mile is a spin-off of the Robotic Systems Lab at ETH Zurich and has already had success integrating artificial neural networks into quadrupedal robots. The company has developed robots that can walk, drive, stand upright on two legs, and manipulate packages. It is now focusing on developing an AI framework that can bring any robotic system to life.

The approach is to integrate reinforcement and supervised learning to enable robots to learn from scratch and adapt their behaviours based on real-world conditions. By continually refining robotic behaviour through data collected from both the real world and artificial scenarios, Swiss-Mile can very rapidly train its robots for new tasks and situations.

The company’s first product applied AI to a robot with legs and wheels, making the robot five times faster, and with ten times the operational range. Swiss-Mile CEO and Co-Founder Dr. Marko Bjelonic told Springwise that what truly sets Swiss-Mile apart is “our high ambition to become the world’s leading robotics company. With this drive, we expanded globally from the very start, even when we were just a team of five.”

The company has recently announced a €19.8 million seed funding round led by Jeff Bezos’ Bezos Expeditions and with participation from the Amazon Industrial Innovation Fund. Bjelonic told Springwise: “With this new investment, we will focus on hiring top talent in AI and closely collaborate with our beta customers to drive a new wave of robotics automation using highly mobile robots.”

Written By: Lisa Magloff