February 17, 2021
Facebook AI is excited to launch the third Habitat Challenge, an open research initiative that invites AI experts around the world to teach machines to navigate real-world environments. The Habitat Challenge 2021, implemented in collaboration with Georgia Tech, asks participants to train embodied agents to perform PointGoal navigation (“Go 5 meters north, 3 meters west”) and ObjectGoal navigation (“Go find a chair”) using Habitat-Sim, Facebook AI’s flexible, high-performance open source 3D simulator.
Habitat Challenge launches at the 2021 Embodied AI Workshop at the Conference on Computer Vision and Pattern Recognition (CVPR), in coordination with eight other embodied AI challenges supported by 15 academic and research organizations. Three of these research competitions will also be based in Habitat-Sim, supported by Facebook AI researchers and our close collaborators. The SoundSpaces Challenge, supported by Facebook AI research scientist Kristen Grauman, Changan Chen (University of Texas at Austin), and Unnat Jain (University of Illinois at Urbana-Champaign), builds on their recent paper and calls for participants to train virtual robots to navigate to audio sources with auditory and visual perception in multi-room 3D environments. Similarly, the MultiON (Multi-Object Navigation) Challenge, hosted by the Indian Institute of Technology Kanpur, the University of Illinois, and Simon Fraser University asks participants to train agents to efficiently navigate to a sequence of objects in a home environment. The Room-Across-Room Habitat Challenge (RxR-Habitat), hosted by Oregon State University, Google, and Facebook AI, builds on object navigation tasks by asking agents to follow human-generated instructions (“Turn left at the corner and go to the kitchen”).
The joint launch of these challenges this year offers the embodied AI research community an unprecedented opportunity to move toward a common framework for the field, converging around a unified set of tasks, simulation platforms, and 3D assets. The organizers will collectively share results across all these challenges at CVPR in June, providing a unique viewpoint on the state of embodied AI research and new directions for the subfield.
One of the central AI research challenges today is teaching machines to move through and operate intelligently in complex situations in the physical world. The potential benefits of this work range beyond conveniences, such as asking a robot to get a set of keys from the kitchen or a laptop from a desk upstairs. Embodied AI can also, for example, help the visually impaired navigate unfamiliar environments or perform difficult tasks in dangerous or difficult situations. AI Habitat is a central component in achieving these goals, with a fast, photo-realistic simulator for embodied research, with an open, modular design that’s both powerful and flexible enough to bring reproducibility and standardized benchmarks to this subfield.
This year, the Habitat Challenge PointGoal navigation task places an agent at a random starting position and orientation in an unseen environment and asks it to navigate to target coordinates. No ground-truth map is available, and the agent must navigate using only sensory input from an RGB-D camera. In ObjectGoal navigation, an agent must start from a random position and orientation in an unseen environment and then find a particular type of object category, such as a table or chair. As with the other challenge, the agent has no map and must use only its sensory input to navigate.
Please review the submission guidelines before entering and note that participants must submit their submissions to EvalAI. The winning team from each track will be invited to nominate a team member to share their work at a CVPR 2021 virtual event, where we will also share the challenge leaderboards.
The iGibson Challenge, hosted by Stanford Vision and Learning Lab and Robotics at Google
Habitat Challenge 2021, hosted by Facebook AI Research (FAIR) and Georgia Tech
Navigation and Rearrangement in AI2-THOR, hosted by the Allen Institute for AI
ALFRED: Interpreting Grounded Instructions for Everyday Tasks, hosted by the University of Washington, Carnegie Mellon University, the Allen Institute for AI, and the University of Southern California
Room-Across-Room Habitat Challenge (RxR-Habitat), hosted by Oregon State University, Google, and Facebook AI
SoundSpaces Challenge, hosted by the University of Texas at Austin and the University of Illinois at Urbana-Champaign
TDW-Transport, hosted by the Massachusetts Institute of Technology
Robotic Vision Scene Understanding, hosted by the Australian Centre for Robotic Vision in association with the Queensland University of Technology Centre for Robotics
MultiON: Multi-Object Navigation, hosted by the Indian Institute of Technology Kanpur, the University of Illinois at Urbana-Champaign, and Simon Fraser University