Bipedal robots have proved a challenging frontier for roboticists, with styles ranging from clunkers to lurchers to those seemingly falling over drunk. However, the AMBER lab at Texas A&M University has created universal mathematical functions of walking derived from human data and optimized for robotic systems. Their own proof of concept robots have strikingly human gaits and react appropriately to disturbances. Furthermore, the system has the potential to be applied to other bipedal robots to similarly upgrade their stride.
The AMBER lab, short for A&M Bipedal Experimental Robotics, recently developed its newest robot named AMBER 2. It can walk stably in two dimensions and is laterally supported by a boom in the third dimension.
UPGRADE TO NEW ATLAS PLUS
More than 1,500 New Atlas Plus subscribers directly support our journalism, and get access to our premium ad-free site and email newsletter. Join them for just US$19 a year.UPGRADE
The AMBER system is fundamentally derived from measuring human parameters in various motions, such as the velocity a knee moves while climbing stairs. However, in this particular example human knees have spikes of rotational velocity that aren't feasible within robotic systems, presenting constraints.
From the data, the team formed a mathematical model of walking which establishes every joint as a pendulum with a corresponding equation. This creates the "canon" of bipedal locomotion. The universal mathematical functions of bipedal locomotion can then be optimized for the known constraints of robotic hardware.
The mathematical solutions that have been developed to marry human data with robotic systems are just as important as the physical hardware developed to test the models and can even be applied to other bipedal robots. In fact, the team uses the commercially-available NAO robot to test new ideas, though the motion isn't as realistic as with the lab's fully articulated Amber.
In disturbance tests, AMBER 2 proves the power of the mathematical modeling, recovering from being pushed or having its “shins” hit with wooden planks. It reacts similarly as a human might, moving a leg backwards to absorb extra momentum or taking an extra compensatory half step forward.
Bipedal locomotion is obviously challenging and mathematically complex to implement, especially compared to other locomotion methods available to roboticists. Our centers of gravity are actually off-center, and we essentially pitch forward as we step, only to catch ourselves with our next step.
So why should robots follow our example? One reason comes from the AMBER lab itself, which applies its research to developing human prosthetic limbs. A more human leg for robots will lead to better robotic legs for humans.
In the video below you can compare a human and AMBER 2 walking side-by-side, and experience schadenfreude as it is pushed over to demonstrate that the boom is only providing lateral stability.
Source: AMBER lab