From the perspective of mechanics, a payload can be described simply as the carrying capacity of a machine (be it a vehicle or a robot), and this capacity is usually measured in terms of weight. In fact, for machines like robots, this capacity is more defined – with exact numbers matching up with the robot’s ability to carry a weight. On the other hand, humans also have their respective payloads, but they can ‘cheat’ their way through such numbers by employing intelligent tactics to overcome the weight. For example, one of us can move a very heavy object by dropping the entire mass of this object on the ground, and then overcoming friction to shove it along the (ground) surface. In other words, while being limited by our payloads, humans also have the advantage of ‘improvisation’ that entails the innate calculation of the weight in question, along with the body balance and posture required to move it.
So, the question arises – what if robots can be endowed with such similar levels of autonomous intelligence when it comes to moving weights? Well, an extensive study done by the researchers from University of Tokyo’s JSK Laboratory has dabbled with this possibility, with their paper being showcased at the ICRA 2015. And their accompanying treat was the modified HRP-2 robot – with its ability to pre-calculate the different ways by which a heavy weight can be tackled.
To give an example – in ‘human terms’, a weight can be tacked and hauled in various manners. We can start by making use of our shoulders to push the weight, or make use of our hips if the mass is higher. Finally, we can completely alter our posture and balance, by leaning against the even-greater weight, and then pushing it with the aid of our legs. The aforementioned HRP-2 does these very same things, by having the ability to autonomously compute the strategies and forces required for moving the weight.
Lastly and more interestingly, the HRP-2 does one better (than humans) by also having the capability to maintain its balance during such rigorous scenarios. According to the scientists, this is done by self-detecting its own tilt, and then modifying its footsteps in accordance with the body gradient.
Images Credit: University of Tokyo / JSK Laboratory