Facebook has teamed up with Carnegie Mellon University to develop PyRobot, a robot control framework that it hopes will allow researchers to boot up and get a robot up and running in a matter of hours, without requiring detailed knowledge of hardware or device drivers. Facebook notes that it wants to provide a robot framework like PyTorch, a deep learning development framework, that provides a level of abstraction to simplify system building and make sharing libraries and tools easier.
PyRobot is a lightweight high-level interface on the ROS robot operating system. It provides a set of hardware independent mediation apis for developers to control various robots. PyRobot abstracts the details of the communication between low-level controllers and programs, so that artificial intelligence researchers no longer need to understand low-level robot operations. Can focus on building advanced ai robot applications.
Researchers can use common features in PyRobot that are applicable to a variety of robots to control joint positions, velocities, or torques, as well as complex functions such as Cartesian path planning or visual SLAM. PyRobot currently supports only LoCoBot and Sawyer, but will continue to support a variety of different robots. While PyRobot provides abstract higher-order control, the researchers can still use different levels of components, such as the ability to bypass the schemer and directly set joint velocities and torques.
Facebook is already using PyRobot for a variety of robot applications, such as point-to-point navigation or push and grab tasks, as well as for remote operations to collect data to train robots. PyRobot includes some off-the-shelf algorithm implementations and provides a way to easily deploy homegrown algorithms to robots. Facebook also mentions that researchers can use PyTorch to train deep learning models and use PyRobot to execute algorithms on robots.
PyRobot can be easier to use robots for research community data set, the algorithm implementation and model, at the same time also can help them to set a benchmark, to compare results, or based on the results of other people moves forward, facebook, said such as low cost in using LoCoBot robot platform, PyRobot helps to reduce the entry barriers, And make it possible to share the results with others. Along with PyRobot’s release, Facebook opened a request for proposals, allowing any research team to submit a PyRobot research proposal paired with LoCoBot. The winner will win a LoCoBot kit for research.