Robotics

Teaching Robots to Learn Teaches the Students Too

• By Dian Schaffhauser
• 08/22/17

Typically, students work with robots that have been pre-programmed or program robots to undertake simple tasks for which the outcome is known. But a research project in Israel came up with a way for high schoolers and first-year engineering students to learn robot intelligence technologies by engaging them in teaching robots — both physical and digital — to learn.

In a paper recently published by the International Journal of Online Engineering, three researchers described how students taught their robots to acquire skills by implementing a "reinforcement learning (RL) process" that used simulation modeling and cloud communication. The idea of RL is to use trial and error rather than direct instructions to help the robot determine appropriate performance criteria — in this case, what angle it should situate itself in to lift varying weights.

The project followed three phases. In the first phase, first- and second-year students doing summer internship projects performed an RL activity in which a humanoid robot learned to adapt its body tilt angle for lifting different weights. That took place at the Center for Robotics and Digital Technology Education at Technion Israel Institute of Technology. In the second phase 11th graders in a local engineering course constructed "animal-like robots" and implemented various RL scenarios that used the approach tested in the first phase. For the third phase university students applied RL, 3D modeling and cloud-based operations to run through a project in which a humanoid robot learned to manipulate multiple joints while maintaining its stability.

Students constructed the elements of the experiment using the ROBOTIS Premium kit, Creo Parametric 3D modeling software from PTC and the Internet of Things development platform from ThingWorx. The cloud platform was used to accumulate, store and process data generated during the robot trials and to transmit data to the robot.

The robot was given an unknown weight while in a seated position. The mass of the weight was estimated by measuring the angular velocity of the robot shoulder joints. Then the robot performed weightlifting trials for different values of body tilt angles, each time attempting to stand up. The robot evaluated whether it succeeded or failed (remained standing or "toppled over") based on data fed to its accelerometer. That data was fed to a local computer via Bluetooth supported with a Python script. Based on the results, the computer would tell the robot what tilt angle was appropriate for the weight being hefted.

The students used the virtual robots to test robot behavior before implementing tasks on the real robots. The benefit of using robotics simulation in a virtual environment, the researchers explained, is that it "allows experimental data to be generated faster, more easily and in any desired quantity, thus significantly [speeding up] the learning process."

The project had two main outcomes. First, researchers found that the development of "learning" robots are highly engaging to students, who gain an understanding of machine learning, parametric design, digital prototyping and simulation, connectivity and the internet of things. Second, they learned that those concepts and technologies are well within the grasp of understanding for high school and first-year college students.

The paper, "Robot Online Learning to Lift Weights: A Way to Expose Students to Robotics and Intelligent Technologies," is available on the iJOE website here.