I’m currently converting my standalone application to a gym vec env. I used the cartpole gym example. However the robot that I’m importing have a camera. But I have no idea how I can get all cameras, is it even possible to fetch all cameras. And then get all the camera data from every camera? I understand it as I’m not allowed to loop in the code and still only run on the GPU?
This is how I create all the robots
car_asset_path = "/home/oscsa287/Desktop/robot/bil1002.usd"
asset_path = "/home/oscsa287/Desktop/robot/bana1.usd"
add_reference_to_stage(usd_path=asset_path, prim_path="/World/world")
create_prim(prim_path="/World/car100", prim_type="Xform", position=self._car_position)
add_reference_to_stage(car_asset_path, "/World/car100")
# create an ArticulationView wrapper for our cartpole - this can be extended towards accessing multiple cartpoles
self._cars = ArticulationView(prim_paths_expr="/World/car100/*")#, name="car_view")
# add Cartpole ArticulationView and ground plane to the Scene
scene.add(self._cars)
# maybe not right
#self.camera = Camera(prim_path="/World/car100/chassi/camera", resolution=self.camera_resolution)
#self.camera.initialize()
#self.camera.add_motion_vectors_to_frame()