Facebook Inc.’s artificial intelligence research team today open-sourced a new robotics framework called PyRobot.

Developed alongside researchers from Carnegie Mellon University, PyRobot is meant to help AI researchers and students integrate deep learning models created using the PyTorch framework with the robots they build. The idea is that they can get their robots up and running with AI skills such as natural language processing more easily.

The company said it wants to help foster “long-term robotics research” to help develop “embodied AI systems” that can learn more effectively through interactions with the physical world.

“PyRobot is a lightweight, high-level interface on top of the Robot Operating System,” Abhinav Gupta, research manager, and Saurabh Gupta, research scientist at Facebook, explained in a blog post. “It provides a consistent set of hardware-independent midlevel APIs [application programming interfaces] to control different robots. PyRobot abstracts away details about low-level controllers and interprocess communication, so machine learning experts and others can simply focus on building high-level AI robotics applications.”

Facebook said PyRobot has dozens of potential applications, for example helping researchers to share data and set benchmarks and build upon one another’s work. The company has requested proposals from the wider AI research community for ideas on how to democratize robotics using PyRobot and LoCoBot, which is a hardware specification and toolkit for building low-cost robots.

PyRobot works by using APIs to abstract functions that robots needs to move around and perform tasks, such as kinematics, path planning, position, velocity and torque control for joints, and visual simultaneous localization and mapping. It comes with a number of pretrained deep learning models that enable robots to navigate, grasp objects and so on. What this means is that developers can program their robots using just a few lines of Python code, Facebook said.

“The cost of hardware and the complexity of specialized software has limited the scale of robotics research,” Facebook’s AI researchers said. “With lower barriers to entry, researchers can deploy several robots that collect data and learn in parallel, for example. By providing a common framework across different hardware, PyRobot will lead to development of benchmarks in robotics (similar to other areas in AI) and quantify the pace of progress.”

Analyst Holger Mueller of Constellation Research Inc. said the release of PyRobot is important because robotics in general is a big part of the wider battle for dominance in machine learning and AI.

“The PyTorch platform was missing a robotics framework, but that’s no longer the case with Facebook open-sourcing PyRobot with promises that it will be easy to use,” Mueller said. “The focus on education is smart as it will make sure future generations of AI developers learn the framework early, and that always gives a leg up in familiarity for the selection of AI / ML / Robotics platforms later in the career of current students.”

Image: Facebook