What is BEHAVIOR-100?

BEHAVIOR-100 is the first generation of BEHAVIOR, a benchmark for embodied AI with 100 activities in simulation, spanning a range of everyday household chores such as cleaning, maintenance, and food preparation. These activities are designed to be realistic, diverse and complex, aiming to reproduce the challenges that agents must face in the real world.


Building blocks of BEHAVIOR-100

Building BEHAVIOR-100 poses three fundamental difficulties for each activity: definition, instantiation in a simulator, and evaluation. BEHAVIOR addresses these with three building blocks. First, we propose a predicate logic-based description language (BDDL) for expressing an activity’s initial and goal conditions, enabling generation of diverse instances for any activity. Second, we identify the simulator-agnostic features required by an underlying environment to support BEHAVIOR-100, and demonstrate in one such simulator, i.e., iGibson 2.0. Third, we introduce a set of metrics to measure task progress and efficiency, absolute and relative to human demonstrators. We include 500 human demonstrations in virtual reality (VR) to serve as the human ground truth.

Do you want to benchmark your solution? Follow the instructions here to get started. The main components are:

* BEHAVIOR-100 benchmark codebase and documentation.
* iGibson simulator codebase and documentation.
* Combined BEHAVIOR-100 iGibson2.0 scene and object assets.
* BDDL specification language codebase and documentation.
* BEHAVIOR-100 VR human demonstration dataset.

What makes BEHAVIOR-100 different?

100 Household Activities in Realistically Simulated Homes

Activities include cleaning, preparing food, tidying, polishing, installing elements, etc. The activities ar eobtained from the American Time Use Survey and approximate the real distribution of tasks performed by humans in their everyday lives.

Activity list | Activity images and videos

Decision Making based on Onboard Sensing for Navigation and Manipulation

These long-horizon activities require that the robot understand the scene, plan a strategy, and execute it by controlling the motion of the embodied agent based on observations. We provide three embodiments, both a rich actuation space and a set of action primitives, and realistic RGB-D and proprioceptive signals: as close as it gets to the challenges of real-world.

Benchmark documentation

More Complex Interactions than just Pick-and-Place

Accomplishing the BEHAVIOR activities requires changing more than the position of the objects in the environment: they need to be cooked, frozen, soaked, cleaned, and more. All these new types of state changes are supported by the provided simulator, iGibson 2.0, and enable new and unique types of activities.

More about the simulator iGibson 2.0