Are you among those people who find getting dressed a very boring task of the day? But guess what? We are here to give you a mechanical technique which will be helping you in getting dressed. Scientists from a renowned institution from Georgia, the process of getting ready is very complicated and involves different physical interactions between the character and his process of clothing which is primarily guided with the person’s touch.
The team of the scientists formulated a technique of dressing by breaking it into small pieces with definitive goals. It lets the person try the give task multiple times and also provides it with the signals when the person tries some useful changes to the policy.
The scientists’ method upgrades the network one step at a time to make the required changes which have the chances of happening in the near future. One of the scientists if the organization said, “We've opened the door to a new way of animating multi-step interaction tasks in complex environments using reinforcement learning. There is still plenty of work to be done continuing down this path, allowing the simulation to provide experience and practice for task training in a virtual world. ”
In the study, the researchers have discovered their method on numerous dressing tasks, like putting on a t-shirt, putting on a jacket and dressing a sleeve with the help of a robot.
With the neural network, they were successfully able to achieve the complicated re-enactment of different ways, an animated character putting on clothes. The key is to include the sense of touch into their framework so as to overcome the challenges in cloth simulation.
The scientists discovered that the careful selection of the cloth examinees and the reward functions in their trained network are important for the success of the framework. As a result, this novel approach not only enables single dressing sequences but a character controller that can successfully dress under various conditions. This entire team of scientist who has discovered this method will be presenting their work at SIGGRAPH Asia 2018 in Tokyo, Japan.