MIT’s ‘Blind’ Cheetah 3 Robot Now Ready To Explore Dangerous Territories


Engineers at Massachusetts Institute of Technology announce the latest improvements in the algorithm and design of the Cheetah robot they have been developing since 2014.

The robot is now ready to explore disaster zones even at zero visibility, without cameras, or any environmental sensors. The Cheetah 3 robot is essentially blind but is programmed to leap over obstacles, gallop over debris, and climb littered stairs. It can even repel external forces that will try to wreck its balance while it performs its task.

Blind Locomotion

The mechanical beast measures 90 pounds and is about the size of an adult Labrador. It will only rely on tactile information and will feel its environment through what the engineers called as "blind locomotion."

To simply explain its new function, Sangbae Kim, the robot's designer and an associate professor of mechanical engineering at MIT, compared the Cheetah 3 robot to a human walking with closed eyes.

"If humans close our eyes and make a step, we have a mental model for where the ground might be and can prepare for it. But we also rely on the feel of touch of the ground," he said.

The "blind locomotion" is made possible by two new algorithms that Kim's team created.

Contact Detection Algorithm

This algorithm helped the robot calculate whether one of his legs will switch from swinging in the air to walking again. Essentially, the robot should know when to stop a leg when it is about to step on an object, to swing that leg in the air, to check whether the object is stable to carry its weight, and to proceed with walking when it predicts that the object will not disrupt its balance.

Each time the robot switch from swinging its feet in the air and stepping on the ground again, the algorithm calculates three probabilities: the probability of a leg making contact with the ground, the probability of the force generated once the leg hits the ground, and the probability that the leg will be in midswing.

The algorithm can even include in its calculations the leg's angle and distance from the ground.

Model-Predictive Control Algorithm

This algorithm calculates how much force a leg should apply once its balance was disrupted by an external force, a force like kicking the robot while it runs. The algorithm will perform these calculations for each leg every 50 milliseconds, or 20 times per second.

Kim explains the algorithm can apply a force in the opposite direction to kill the unwanted velocity that tried to put it off-balance.

"If I apply 100 newtons in this opposite direction, what will happen a half second later?" Kim said, demonstrating how this algorithm processes its data.

Cheetah 3 Robot Developments

In 2015, Kim and his team of engineers have programmed the robot to jump over hurdles while it is running. At this time, the Cheetah 3 robot became the first four-legged robot to run and jump over obstacle autonomously.

The robot was able to jump over obstacles of about 18 inches tall, which is more than half of the robot's height while maintaining an average running speed of 5 miles per hour.

In 2014, the engineers were able to make the robot sprint to up to 10 miles per hour and to run and jump untethered across the grass. Recently, the engineers estimated the robot may eventually attain a running speed of up to 30 miles per hour.

ⓒ 2018 All rights reserved. Do not reproduce without permission.
Real Time Analytics