There comes a point in time for all of us where the only option to find a thing hidden beneath the living room couch is by using our sense of touch. But now we are not alone in the matter as researchers are trying on robots to possess the similar capability.
It all started back in 2019 when a team of scientists from the Massachusetts Institute of Technology (MIT) tried to work on the idea by combining tactile sensors and AI all with the hopes of letting robots identify normal objects only by touching them.
Now moving on a more advanced level, there is another separate group of scientists from MIT who have made the similar wish possible with their new machine called RF Grasp.
It is comprised of a wrist-mounted camera and an RF reader which makes it makes it possible for the robot to hone in and then also pick the object. But for the identification part, the object is bound to have an RF tag on it as the robot would only then be able to find it - even if the object is hidden somewhere behind the walls.
Fortunately, this is one of those university projects that will have great practical usage as well since the team is hoping to help companies like Amazon to better automate and streamline their warehouses with RF Grasp. One of the researchers Associate Professor Alberto Rodriguez explained the value of it really well by stating how perception and picking currently stand as two roadblocks in the industry.
According to the team, the most challenging part of creating RF Grasp was to integrate both sight and RF vision for a smooth decision-making process. The researchers have trained the system in place by matching it to how we would react to a sound being played at a far distance. We would of course turn our head to see its source.
While RF Grasp will rely on the RF reader to find tagged objects, but it will also collect more information through its camera to make the final decision regarding picking up an object.
When compared to the traditional robots with visual systems, RF Grasp can first locate and then also pick up objects in almost half total movements. The robot can also clean up and declutter the working place while also performing the given task.
Some may argue that tagging every important object with RF tags can be a bit too much but we don’t see it as a potential barrier. In fact, in countries like Japan, there is already a trend of deploying RF tracking tags in the retail sector.
So, precisely the technology already supports RF Tags.
It all started back in 2019 when a team of scientists from the Massachusetts Institute of Technology (MIT) tried to work on the idea by combining tactile sensors and AI all with the hopes of letting robots identify normal objects only by touching them.
Now moving on a more advanced level, there is another separate group of scientists from MIT who have made the similar wish possible with their new machine called RF Grasp.
It is comprised of a wrist-mounted camera and an RF reader which makes it makes it possible for the robot to hone in and then also pick the object. But for the identification part, the object is bound to have an RF tag on it as the robot would only then be able to find it - even if the object is hidden somewhere behind the walls.
Fortunately, this is one of those university projects that will have great practical usage as well since the team is hoping to help companies like Amazon to better automate and streamline their warehouses with RF Grasp. One of the researchers Associate Professor Alberto Rodriguez explained the value of it really well by stating how perception and picking currently stand as two roadblocks in the industry.
According to the team, the most challenging part of creating RF Grasp was to integrate both sight and RF vision for a smooth decision-making process. The researchers have trained the system in place by matching it to how we would react to a sound being played at a far distance. We would of course turn our head to see its source.
While RF Grasp will rely on the RF reader to find tagged objects, but it will also collect more information through its camera to make the final decision regarding picking up an object.
When compared to the traditional robots with visual systems, RF Grasp can first locate and then also pick up objects in almost half total movements. The robot can also clean up and declutter the working place while also performing the given task.
Some may argue that tagging every important object with RF tags can be a bit too much but we don’t see it as a potential barrier. In fact, in countries like Japan, there is already a trend of deploying RF tracking tags in the retail sector.
So, precisely the technology already supports RF Tags.
