Virtual Reality

NVIDIA Project Holodeck Trains Robots and AI with Virtual Reality

Share

Computer graphics conference SIGGRAPH 2017 was a memorable event: for the first time ever, participants could see a robot playing dominoes in a virtual reality environment. The robot was trained through the Isaac deep learning AI training program running on the Holodeck virtual reality platform developed by NVIDIA.

 

A Promising Experiment

The Holodeck VR platform is still in the experimental phase. It was created with the purpose of allowing users to view real life objects rendered in virtual reality in high detail. In addition to rendering, Holodeck can also replicate the laws of physics with regard to interactions between objects, people and the environment.

Alongside the Holodeck platform, NVIDIA revealed a breakthrough program for training artificial intelligence entities, named Isaac. The next logical step was to bring together these new technologies and create an advanced robot training virtual reality platform.

 

Virtual Reality and Robots– a Successful Collaboration

Using Isaac and Holodeck, NVIDIA specialists managed to upload the gameplay algorithm into the robotic neural network and then connect it to the virtual reality environment. Thus, the robot trained by the Isaac program was able to engage in a game of dominoes with a human player.

The keynote presentation held by NVIDIA CEO and founder Jensen Huang presents the Isaac robot simulator playing golf, showcasing the fast and accurate integration of physics and high definition graphics:

 

What Is the Next Move?

Programming robots to interact with humans in a virtual reality environment is just the first step towards, potentially, a new industrial revolution. Until now, humans have only created, programmed and controlled robots in real life.

As new technologies, especially virtual reality, have become a part of our daily life and business, there is a growing need for more precise and advanced integration of VR into various fields of industry.

After all, if robots are capable of building cars using the design and a set of instructions, isn’t it possible to bring them onboard the design phase, where engineers already work with virtual reality headsets and apps?

For now, NVIDIA has proven that robots are very easy to train in an immersive environment. Machine learning is sufficiently advanced to allow artificial intelligence entities to integrate large quantities of data in a short time. And, unlike humans, robots do not get motion sickness and do not experience any of the negative side effects of being immersed in a virtual reality environment for too long.

Right now, robots can play dominoes with people in VR. Soon enough, they may be able to work side by side with human specialists, to create better products and services.

If you are interested in the Holodeck platform, you can contact NVIDIA directly at the bottom of their official web page.

ARPost

Recent Posts

How Virtual Reality Is Revolutionizing Police Training

Virtual reality creates safe and immersive environments that help police officers hone their skills, equipping them with new ways to…

1 year ago

Hands-On Review: YOGES Handle Attachments for Quest 2 Controllers

YOGES has an innovative take on adapters for Quest 2 controllers, including a unique removable piece that turns both controllers…

1 year ago

Exploring the World of Live XR Theater

Live XR theater started during the pandemic, but it’s more than a way to avoid a crowd. The medium allows…

1 year ago

Challenges Behind Applying Real-World Laws to XR Spaces and Ensuring User Safety

Applying real-world laws to XR spaces will require governments, businesses, and institutions to work together for effective regulation.

1 year ago

Alien Invasion AR FPS Review

Alien Invasion AR FPS does a lot with relatively little as it hooks me into the story through a combination…

1 year ago

Talespin Launches AI Lab for Product and Implementation Development

AI is like any tool - it makes the job easier but only if you know how to use it.…

1 year ago