During the first phase of development, we mainly tested the framework with toy environments such as the CartPole and the Pendulum.
For the first stable release, we want to achieve simulating complex multi-DoF floating-base robots, whose models might contain meshes.
In #4, we included models of the iCub humanoid robot, but after the first experiments we never used them for real development. We should start again that activity and finalize their usage in simulated environments.
We can use this issue to report the problems and bottlenecks we encounter. To begin:
Simulating iCubGazeboV2_5 (model, world) is extremely slow. I get a RTF ~ 4%. We should run a profiler (#91) to understand where is the bottleneck (Is the problem DART or Gazebo?).
Note that the model still has rendering problems, as reported in #31.
During the first phase of development, we mainly tested the framework with toy environments such as the CartPole and the Pendulum.
For the first stable release, we want to achieve simulating complex multi-DoF floating-base robots, whose models might contain meshes.
In #4, we included models of the iCub humanoid robot, but after the first experiments we never used them for real development. We should start again that activity and finalize their usage in simulated environments.
We can use this issue to report the problems and bottlenecks we encounter. To begin:
iCubGazeboV2_5
(model, world) is extremely slow. I get a RTF ~ 4%. We should run a profiler (#91) to understand where is the bottleneck (Is the problem DART or Gazebo?).Note that the model still has rendering problems, as reported in #31.