OffWorld Gym: open-access physical robotics environment for real-world reinforcement learning benchmark and research

Abstract

Success stories of applied machine learning can be traced back to the datasets and environments that were put forward as challenges for the community. The challenge that the community sets as a benchmark is usually the challenge that the community eventually solves. The ultimate challenge of reinforcement learning research is to train real agents to operate in the real environment, but until now there has not been a common real-world RL benchmark. In this work, we present a prototype real-world environment from OffWorld Gym -a collection of real-world environments for reinforcement learning in robotics with free public remote access. Close integration into existing ecosystem allows the community to start using OffWorld Gym without any prior experience in robotics and takes away the burden of managing a physical robotics system, abstracting it under a familiar API. We introduce a navigation task, where a robot has to reach a visual beacon on an uneven terrain using only the camera input and provide baseline results in both the real environment and the simulated replica. To start training, visit https://gym.offworld.ai.

Train Offline, Test Online: A Real Robot Learning Benchmark

“…Georgia Tech's Robotarium [17] allows for remote experimentation of multi-agent methods on a physical robotic swarm, which has been extensively used not just in research but also in education. OffWorld Gym [18] provides remote access to navigation tasks using a mobile robot with closely mirrored simulated and physical instances of the same environment. A recent survey paper [19] provides an overview of robotic grasping and manipulation competitions, including some involving remotely-accessible, shared robots such as [20].…”

Reinforcement-Learning-Based Routing and Resource Management for Internet of Things Environments: Theoretical Perspective and Challenges

“…In these cases, the agent must choose one of the available actions at each step. On the other hand, in other environments, such as controlling a robot in a physical world, the action space is continuous [58]. This means the agent can choose an action from a continuous range of values rather than a limited set of options.…”

Uncrewed Ground Vehicles (UGVs) and Nature-Inspired Designed Robot DIGIT and SPOT: A Review

“…A challenge of Reinforcement Learning (RL) is training real agents to operate the robot in a real-world environment. Kumar et al (2019) developed a collection of open-access real-world environments named Off World Gym to solve RL's robotics issue. A Husarion Rosbot was also used for the experiment of the API interface, which shows the seamless transition between the real world and the simulation environment (Kumar et al, 2019).…”