Tech

Watch this sightless robot learn how to walk just like a human

Researchers are trying to teach robots that walking is more than just simply putting one foot in front of the other.

Learning how to walk is difficult. It’s even more difficult if you don’t have the benefit of millions of years of human evolution to help you out.

Shutterstock

That’s exactly why researchers are training our robotic counterparts how to use every tool in the locomotion toolbox to help keep themselves upright and get them walking straight and hazard free.

In a recent demonstration, researchers from the Technical University of Munich (TUM) in Germany showed off some new tricks from their humanoid robot, LOLA, including something they call multi-contact locomotion.

LOLA the humanoid robot.

Using an experimental software, LOLA is able to sense its environment without the help of cameras. Instead, the bot uses various sensors to recognize uneven surfaces, pushing, and loose terrain to help it adapt to environmental factors in real-time.

“For these experiments the navigation module is disabled, thus the foothold positions and hand contact points are manually set according to the current environment setup. However, disturbances (pushing, uneven terrain, rolling board) are not known to the robot and are compensated by our online stabilization methods.”

While teaching robots how to walk may, on the surface, seem like a pretty basic step toward functional humanoid bots, bi-pedal locomotion has been a difficult nut to crack for roboticists looking to make upright machines that are applicable in the real world.

According to an interview with Institute of Electrical and Electronics Engineers, the researchers plan to combine their multi-contact algorithm with computer vision that can map out rooms and identify traversable areas.

Though fairly advanced among bi-pedal robots, LOLA still has some work to do before it’s capable of real locomotion. For instance, the bot isn’t yet capable of walking autonomously since its navigations systems haven’t yet been completed, meaning all the footfalls seen in the demonstration had to be pre-programmed.

Researchers also hope to give LOLA more reactive skills in addition to proactive locomotion, meaning it could potentially use its multi-contact software to stabilize quickly in situations where it’s suddenly been thrown off balance.

Eventually, researchers say they’d like to increase LOLA’s walking speed to that of a human, though it’s always better to learn how to walk before you start to run.

Hey! Not bad.

Share