Header

Robotic Expectations

Insights into assistive robots, autonomous drones, and a demo of the humanoid robot Unitree G1 were some of the highlights of an event co-organised by ZHAW digital, the University of Zurich, Lab42, and Mindfire. At two locations, more than 200 visitors watched the robot perform live and heard about research projects and prospects in Zurich and Winterthur.

Impressions from the event “Flexibility in action” in Winterthur at the ZHAW and in Oerlikon at the University of Zurich. Photos by Lars Neumayer.
Impressions from the event “Flexibility in action” in Winterthur at the ZHAW and in Oerlikon at the University of Zurich. Photos by Lars Neumayer.
Impressions from the event “Flexibility in action” in Winterthur at the ZHAW and in Oerlikon at the University of Zurich. Photos by Lars Neumayer.
Impressions from the event “Flexibility in action” in Winterthur at the ZHAW and in Oerlikon at the University of Zurich. Photos by Lars Neumayer.
Impressions from the event “Flexibility in action” in Winterthur at the ZHAW and in Oerlikon at the University of Zurich. Photos by Lars Neumayer.

The Unitree G1 is one of several humanoid robots being developed worldwide, mainly in China and the US. The model is one of the first that can be purchased for $16,000 – with a certain set of functions; it becomes more expensive when one wants to add hands, for example. In the demo, the robot was controlled by a human and able to stand up, walk, wave, shake hands, and regain balance after being pushed (except for one side-push).

Kaltura Privacy Warning

Kaltura prevents you from watching videos or listening to sound files anonymously. When you play Kaltura videos or sound files, Kaltura will register it.

Videoclip from the event by Johanna Seiwald

Why do we need humanoid robots?

Yulia Sandamirskaya, who leads the Research Centre Cognitive Computing at the ZHAW Institute of Computational Life Sciences (ICLS), School of Life Sciences and Facility Management, explains why robots are not only shaped like Roombas but also in human form. “The human shape is very versatile; we evolved to be able to solve many complex tasks with our hands, legs etc.” Humanoids are being developed to assist humans, for instance, with their daily chores or when people are injured or elderly. “We still need a lot more research, to put the user first and ensure the robots are safe.”

How do we make robots “smart”?

When Yulia says “smart”, she does not mean intelligent, but rather agile, flexible and able to learn quickly. So far, we have the hardware, as well as good movement control and sensing, but the ability to sense and understand the environment fast enough for real-time motion planning and control is lacking. Additionally, when AI is being implemented it is still very slow, inflexible, and demands a lot of energy. “We need embodied AI to improve the interaction of robots with the real world”, says Yulia. In one of her projects, she utilises neuromorphic computing, which can enable real-time 3D vision for object detection, localisation, and pose estimation – key tasks for vision in robotics.

Flying to the future

In Zurich, Davide Scaramuzza from the Robotics and Perception Group of the University of Zurich showcased a race of autonomous drones. What is special about these drones is that they are not controlled by humans or GPS but are trained with AI, vision-based, and fly at an unprecedented speed. Drones like these are useful not only for drone races or infrastructure inspection but also for fire prevention, water use control etc. While there are many use cases where robots can be used for good, a lot of innovation and development still needs to happen. “We need to bridge the gap between the academic research in robotics and practical applications”, concludes Yulia.

More information:

go back