A saying I hear from many startups is that “there’s no need to rethink the caliper.” It is something that I appreciate from an economic point of view. It’s expensive, resource intensive, and you’re probably better off spending both your time and money elsewhere when there are so many effectors on the market already.
I also recently made a claw machine analogy during an interview, and it got some criticism. Now I understand a little better why this is so, at least in part. Speaking of its new approach to robotic gripping, MIT invokes the perennial arcade favorite, noting: “By manipulating a gaming claw, a gamer can plan whatever he wants. But once you hit the joystick button, it’s a game of wait and see. If the claw misses the mark, you’ll have to start from scratch for another chance to win a prize.”
Image Credits: MIT
If you think about it for a moment, you realize you’re suddenly faced with something that comes up again and again in this field of study: This is not how humans approach work, and there is a reason for it. If, for example, you are grasping an object with an odd or unexpected weight distribution, you usually don’t need to remove your hand and try again. you adjust.
The team describes a system that adjusts to an object in real time, using reflections and feedback. Says MIT:
If the gripper fails to grab the object, instead of backing up and starting over like most grippers do, the team wrote an algorithm that tells the robot to quickly perform any of three grasping maneuvers, which they call “reflections”, in response to real-time measurements at your fingertips. All three reflexes are activated within the last inch of the robot as it approaches an object, allowing the fingers to grasp, pinch or drag an object until it has a better grip.
Interestingly, the project is based on actuators developed for the school’s mini cheetah robot, which were designed to help it react on the fly on uneven terrain. The new system is built around an arm with two multi-jointed fingers. There’s a camera at the base and sensors at the tips that record feedback. The system uses that data to adjust accordingly.
The team is currently using the clamp to clean the lab. Says MIT:
They placed a variety of household objects on a shelf, including a bowl, a mug, a can, an apple, and a bag of ground coffee. They showed that the robot could quickly adapt its grip to the particular shape of each object and, in the case of coffee grounds, to the softness. Out of 117 attempts, the gripper quickly and successfully picked up and placed objects more than 90 percent of the time, without having to go back and start over after a failed grip.
—————————————————-
Source link