Jul. 08, 2024
Mechanical flexibility in robot manipulators is due to compliance at the joints and/or distributed deflection of the links. Dynamic models of the two classes of robots with flexible joints or flexible links are presented, together with control laws addressing the motion tasks of regulation to constant equilibrium states and of asymptotic tracking of output trajectories. Control design for robots with flexible joints takes advantage of the passivity and feedback linearization properties. In robots with flexible links, basic differences arise when controlling the motion at the joint level or at the tip level.
Click here to get more.
This robot is the first full-size humanoid able to rebalance after accidental or intentional shoves or kicks landed anywhere on its body, thanks to new software
(Image: Computational Neuroscience Laboratories, ACT/IEEE)
If robots are going to work alongside humans, then they will need to stand up to accidental bumps and shoves, not to mention the occasional deliberate kick.
That is why researchers in Japan have developed software that allows a life-size humanoid robot to stay on its feet no matter where on its body it is pushed. Theirs is the first full-size humanoid to show such steadiness others of similar size inevitably topple over when nudged in the right spot. In experiments, the robot was subjected to repeated pushes. A virtual robot received much harder shoves.
Rebalancing should allow humans to interact more naturally with robots, letting them act as a physical guide, for example. If a controller tries to show other full-size humanoids how to perform a task by moving its limbs, there is a strong chance the thing will fall over.
Advertisement
The robot, made by US firm Sarcos and then developed by researchers at the National Institute of Information and Communications Technology in Japan, suffers no such unsteadiness, it can easily rebalance when its arms are pulled into different positions.
The robots balancing ability depends on its joints. For one thing they are never kept rigid, even when standing still, meaning they yield slightly when the robot is pushed.
If you want to learn more, please visit our website Fuxin Intelligent.
Force sensors within each joint also work out the position and velocity of the robots centre mass as it moves around. Control software rapidly figures out what forces the robots feet need to exert on the ground to bring it back into balance, and tells the joints how to act.
As well as keeping the robot steady as it moves itself around, the technique lets it readjust to sudden, external forces.
In cases when the robots joints cannot quickly swing its centre of mass back into place, it ends up staggering a bit like a boxer after a heavy punch. This constitutes several rounds of rebalancing, with each cycle shifting the centre of mass closer to its original balance point.
Some other humanoid robots rebalance themselves by measuring changes to the position of each joint. This requires very accurate knowledge of the magnitude of an applied shove, says ATR researcher Sang-Ho Hyon, which is difficult to achieve without covering the whole robot with force sensors.
Most robots lack such sensors, and so use a relatively simple trick to rebalance themselves. For example, Hondas ASIMO, shifts its hip joint in order to stay steady, which only works in some cases.
Imagine what happens if we push ASIMOs hip, says Hyon. This would leave the diminutive humanoid unable to rebalance. In our method, when we push the hip, the hip follows the external forces and other joints compensate for the balance, adds Hyon.
You just dont see the real good humanoids out there get pushed, says Jerry Pratt, a roboticist at the Institute for Human and Machine Cognition (IHMC) in Pensacola, Florida, US.
This team is currently ahead of the pack in terms of having it work on a full robot, Pratt told New Scientist. Making the robot more compliant instead of stiff plays a big part in that, he says, and the ability to measure and control the torque force at every joint is also crucial.
Pratt and colleagues are working on their own control strategy, which involves rebalancing with a single step. Imagine you are crossing a pond and you can only step to one rock to rebalance, he says. The software will be tested next year after the team finishes building a suitable humanoid.
Journal reference: IEEE Transactions on Robotics (vol 23, p884)
Robots Learn more about the robotics revolution in our continually updated special report.
Are you interested in learning more about flexible joint robot? Contact us today to secure an expert consultation!
If you are interested in sending in a Guest Blogger Submission,welcome to write for us!
All Comments ( 0 )