What comes to mind when you think of robotics? Dancing robots from Boston dynamics? Self-driving cars? Or the Roombas and vectors? Whatever it is, the common theme is that the heart that makes them tick is software. Some of the most important problems in robotics are computer science problems and one of the major bottlenecks now, be it in self-driving or service robots, or material handling is in software. Needless to say, writing software for robots is ridiculously hard! Now, let's see why it is hard and what ROS does to make it easier.
Read the rest at my college magazine, aperq
This article is an analysis of the research paper “Grasping in the wild : Learning 6DoF Closed-LoopGrasping from Low-Cost Demonstrations”, Shuran Song Et al. Whoa, thats a mouthful! Don’t worry, we will do it one-by-one! I have tried to introduce all the terminology as easily as possible, so if you dont understand them, please do let me know. Also, do follow the links and research on your own for a much deeper understanding of the topics discussed if this sparked your interest.
So, the problem we are trying to solve is the seemingly simple task of grasping objects using a robotic arm with a binary gripper at the end (it can either be completely closed or completely open, hence binary). As you can imagine, this is a very important task for robots to interact with objects, be it in an industrial setup of in a household.
Read the rest at My medium page