How they work
The first chambers were developed at the Seminar for Statistics of ETH Zurich, initially conceived as testbeds for causal inference algorithms. They were presented in an open-access paper in Nature Machine Intelligence, together with their open-source blueprints and a collection of public datasets.
In a nutshell, a chamber contains a physical system and allows us to control and measure its variables without human supervision. It provides validation tasks with a ground truth for a variety of algorithms from ML, AI, statistics, and engineering.
In principle, any physical system is chamberizable as long as (1) its variables can be digitally measured and controlled, and (2) it can be reset without human supervision.
Because the physical system is well understood, we can also provide a ground truth for causal inference tasks, as well as mechanistic models and simulators of the physical phenomena inside the chambers. These are particularly useful for studying problems in Sim2Real, Simulation-Based Inference, Hybrid Learning, and related fields. See the case studies for examples.
Hardware configurations
A chamber can operate under several configurations, exposing different variables of the underlying physical system. Furthermore, the chamber can set control inputs as functions of other sensor measurements, allowing for feedback mechanisms and tunable causal effects.
Configurations are specified at the beginning of an experiment and loaded automatically by the chamber.
Light Tunnel Mk2: hardware configurations
See the hardware configurations for the Light Tunnel Mk2
Wind Tunnel Mk2: hardware configurations
See the hardware configurations for the Wind Tunnel Mk2
Chamber control language
The chambers are controlled through a simple language with three instructions.
This sets the variable target to the given value, returning when the change has been made in the hardware.
See the hardware configurations of each chamber for a list of variables and their valid and default values.
The chamber returns n successive measurements of all variables, including images if it produces them. Setting delay (in milliseconds) adds a delay between measurements, allowing you to change the measurement frequency.
The maximum measurement frequency, i.e., with delay=0, is approx. 12Hz for the Wind Tunnel Mk2 and 10Hz for the Light Tunnel Mk2.
The chamber acts as a precise clock and waits for the given milliseconds before executing the next instruction.
Instructions can be sent synchronously (see real-time experiments) or as an experiment protocol, which is placed on a queue and executed when a chamber becomes available.
Last updated