This is an old revision of the document!


SCONE Concepts

SCONE Scenario

In SCONE, everything needed to perform a predictive simulation is bundled in what is called a scenario. Each SCONE scenario consists of the following components:

  • A musculoskeletal model of the human or animal, which is used in the simulation.
  • A controller, which generates input for the model actuators.
  • An objective, which describes the goal task for which you wish to optimize, through a weighted combination of measures.
  • An optimizer, which attempts to find the parameters that best perform the specified objective.

SCONE Models

SCONE is designed to work with OpenSim for modeling and simulation. Any type of model is supported (both 2D and 3D), as long as the model contains controllable actuators, such as hill-type muscle-tendon units or torque-based actuators. In case a model needs to interact with the environment, it must also contain collision shapes and contact models. This is needed to compute the reaction forces during the forward dynamics simulation. Finally, models should typically contain limit forces to simulate the forces generated by tendons and bony structures when a joint reaches a certain limit.

Models used for predictive simulations should be as simple as possible. The reason for this is threefold: 1) Models with less degrees-of-freedom are more restricted in their motion, making it easier to find an optimum; 2) The less actuators, the less control parameters are needed during the optimization; and 3) Simulation performance decreases with model complexity, so optimizations will run much quicker with simpler models.

SCONE Controllers

Controllers compute the input values of the model actuators.

In SCONE, controllers consist of modular components that can be combined to perform complex behaviors.

SCONE Objectives and Measures

Measures define the fitness or quality of a simulation. The goal of the optimizer is to minimize (or maximize, depending on the type) the measure.

SCONE Optimizers

  • by tgeijten