NVIDIA and Robotic Systems Lab Develop Robots That Can Climb and Jump

Roboticists from Robotic Systems Lab and NVIDIA Embedded Teach Robots to Climb and Jump

Roboticists from the Robotic Systems Lab at ETH Zürich and NVIDIA Embedded have taught four-legged robots to climb and jump. After training in simulation, the robots can autonomously decide how to scramble over and under obstacles, which will one day help them do dangerous jobs so that humans don't have to.

The robots are equipped with NVIDIA Jetson Nano devices, which power their artificial intelligence (AI) algorithms. The AI algorithms allow the robots to learn how to move their limbs in order to climb and jump. The robots are also equipped with sensors, such as cameras and depth sensors, which help them to perceive their surroundings.

The researchers trained the robots in simulation using the NVIDIA Isaac Sim software platform. Isaac Sim is a physics-based simulation platform that allows robots to be trained in a safe and controlled environment. The researchers used a reinforcement learning approach to train the robots. Reinforcement learning is a type of machine learning in which the robot learns by trial and error. The robot is given a reward for completing a task, and it learns to avoid actions that result in a penalty.

The researchers tested the robots in the real world and found that they were able to climb and jump over obstacles. The robots were also able to avoid obstacles and navigate around a cluttered environment.

The research team believes that this work is a significant step towards developing robots that can perform dangerous or hazardous tasks. For example, the robots could be used to inspect nuclear power plants or to search for survivors in collapsed buildings.

The research was published in the journal Nature Machine Intelligence.

Here are some of the key takeaways from the research:

  • The robots were able to learn how to climb and jump using a reinforcement learning approach.
  • The robots were able to transfer their skills from simulation to the real world.
  • The research is a significant step towards developing robots that can perform dangerous or hazardous tasks.

This research has the potential to revolutionize the field of robotics. By teaching robots how to climb and jump, researchers are opening up new possibilities for how robots can be used. These robots could be used to explore dangerous environments, such as collapsed buildings or nuclear power plants. They could also be used to perform tasks that are difficult or dangerous for humans, such as search and rescue or bomb disposal.

You are currently viewing NVIDIA and Robotic Systems Lab Develop Robots That Can Climb and Jump

Leave a Reply