Skip to content

Upgrading Existing APIs#

There have been several API changes in AutonomySim v1.2 that we hope removes inconsistency, adds future extensibility and presents cleaner interface. Many of these changes are however breaking changes, which means that you will need to modify your client code that communicates with AutonomySim.

A Faster Way#

While most changes you need to do in your client code are fairly easy, a faster way is to simply take a look at the example code, such as Hello Droneor Hello Car, to gain an understanding of the changes.

Importing AutonomySim#

Instead of this:

from AutonomySimClient import *
use this:

import AutonomySim

Above assumes you have installed AutonomySim module using, 

pip install --user AutonomySim

If you are running you code from PythonClient folder in repo then you can also do this:

import setup_path 
import AutonomySim

Here setup_path.py should exist in your folder and it will set the path of AutonomySim package in PythonClient repo folder. All examples in PythonClient folder uses this method.

Using AutonomySim Classes#

As we have everything now in package, you will need to use explicit namespace for AutonomySim classes like shown below.

Instead of this:

client1 = CarClient()

use this:

client1 = AutonomySim.CarClient()

AutonomySim Types#

We have moved all types in AutonomySim namespace.

Instead of this:

image_type = AutonomySimImageType.DepthVis

d = DrivetrainType.MaxDegreeOfFreedom

use this:

image_type = AutonomySim.ImageType.DepthVis

d = AutonomySim.DrivetrainType.MaxDegreeOfFreedom

Getting Images#

Nothing new below, it's just combination of above. Note that all APIs that previously took camera_id, now takes camera_name instead. You can take a look at available cameras here.

Instead of this:

responses = client.simGetImages([ImageRequest(0, AutonomySimImageType.DepthVis)])

use this:

responses = client.simGetImages([AutonomySim.ImageRequest("0", AutonomySim.ImageType.DepthVis)])

Utility Methods#

In earlier version, we provided several utility methods as part of AutonomySimClientBase. These methods are now moved to AutonomySim namespace for more pythonic interface.

Instead of this:

AutonomySimClientBase.write_png(my_path, img_rgba) 

AutonomySimClientBase.wait_key('Press any key')

use this:

AutonomySim.write_png(my_path, img_rgba)

AutonomySim.wait_key('Press any key')

Camera Names#

AutonomySim now uses names to reference cameras instead of index numbers. However to retain backward compatibility, these names are aliased with old index numbers as string.

Instead of this:

client.simGetCameraInfo(0)

use this:

client.simGetCameraInfo("0")

# or

client.simGetCameraInfo("front-center")

Async Methods#

For multirotors, AutonomySim had various methods such as takeoff or moveByVelocityZ that would take long time to complete. All of such methods are now renamed by adding the suffix Async as shown below.

Instead of this:

client.takeoff()

client.moveToPosition(-10, 10, -10, 5)

use this:

client.takeoffAsync().join()

client.moveToPositionAsync(-10, 10, -10, 5).join()

Here .join() is a call on Python's Future class to wait for the async call to complete. You can also choose to do some other computation instead while the call is in progress.

Simulation-Only Methods#

Now we have clear distinction between methods that are only available in simulation from the ones that may be available on actual vehicle. The simulation only methods are prefixed with sim as shown below.

getCollisionInfo()      is renamed to       simGetCollisionInfo()
getCameraInfo()         is renamed to       simGetCameraInfo()
setCameraOrientation()  is renamed to       simSetCameraOrientation()

State Information#

Previously CarState mixed simulation-only information like kinematics_true. Moving forward, CarState will only contain information that can be obtained in real world.

k = car_state.kinematics_true

use this:

k = car_state.kinematics_estimated

# or

k = client.simGetGroundTruthKinematics()