Pick-and-place robots are predicted to fully replace factory floor workers in the future. However, before these robots take over the floor, they will have to become more flexible with the objects they can pick-and-place. It all has to do with the robot’s ‘vision.’
Making pick-and-place robots ‘see’ the objects they are to be picking up is not easy work. Things can often go wrong when a robot tries to pick up an object it has never encountered before.
However, a team from MIT’s Computer Science and Artificial Intelligence Laboratory developed the idea that they could assign 3D key points which would help the robot identify where and how to pick new objects up. They call their innovation kPAM – or Key point Affordance Manipulation.
Source: CSAIL | kPAM
Russ Tedrake, the senior author of the report, wrote:
“Understanding just a little bit more about the object – the location of a few key points – is enough to enable a wide range of useful manipulation tasks..”
The team has successfully managed to get a robot to pick up an object without a fixed template. They say this is done through the novel formulation of ‘category-level manipulation’ using the 3D key points.
“Using this formulation, we factor the manipulation policy into instance segmentation, 3D key point detection, optimization-based robot action planning and local dense-geometry-based action execution. This factorization allows us to leverage advances in these sub-problems and combine them into a general and effective perception-to-action manipulation timelines.”
Their implementation of the system allowed for a robot to be able to pick up a mug and clean a pair of shoes, both without a prior template for the robot to follow.
The team is hoping that kPAM can be useful for more applications than just robot arms in the future. Their research could help transform the domestic robots that one could see in households in the next decades. And they also hope it will continue to assist in the automating of factory floors and the replacing of the more repetitive tasks within them.
Works Cited
“MIT Robot Uses 3D Keypoints for Advanced Coordination.” The Engineer, 18 Mar. 2019, www.theengineer.co.uk/mit-robot-keypoints/?cmpid=tenews_7701303&utm_medium=email&utm_source=newsletter&utm_campaign=tenews&adg=5A0EAC12-A7C5-4FD9-9AA8-84D6979695B5.
“KPAM.” Google Sites, sites.google.com/view/kpam.
