REAL WORLD ROBOTICS

The Presentation and Hardware

The first part of the final challenge focuses on presentation and hardware. The presentation allows the students to explain their work, their clever solutions and the results they obtained. Afterwards, the examiners evaluated the creativity, mechanical cleverness and robustness of the designs.

Software and Reinforcement Learning

Students have 30 minutes to demonstrate their achievements, testing their designs in three teleoperation tasks. One operator per group should control the robot hand with the aid of vision algorithms and mapping and show that he can achieve each one of the following tasks:

• Power grasp
• Drill pick-up and operation
• Fidget spinner operation

Secondly, the team should show dexterity by applying Reinforcement Learning to the task 'Get the ball rolling, rotate a tennis ball inside the palm of your hand in both directions of an x-axis rotation' Finally, students can show more personal achievements that involve traditional control and/or Reinforcement Learning and impress the jury.

Students will showcase in a live demo their solutions to one challenge selected from a variety of options. The challenges include:

• Upgrading the Motors and Actuation Tower
• PIP-DIP Coupling
• Introducing Fingertip Tactile Sensors
• Upgrading the Motors and Actuation Tower
• Using RGB-only input for real-time teleoperation
• Using RL to learn to play Jenga
• Using Gaussian Splatting for enhanced simulations