Skip to main content

Client SDK for InvertedAI

Project description

Documentation Status PyPI Open In Colab

InvertedAI

Overview

Inverted AI provides an API for controlling non-playable characters (NPCs) in autonomous driving simulations, available as either a REST API or a Python SDK, (and C++ SDK) built on top of it. Using the API requires an access key - contact us to get yours. This page describes how to get started quickly. For more in-depth understanding, see the API usage guide, and detailed documentation for the REST API, the Python SDK, and the C++ SDK. To understand the underlying technology and why it's necessary for autonomous driving simulations, visit the Inverted AI website.

Getting started

Installation

For installing the Python package from PyPI:

pip install --upgrade invertedai

The Python client SDK is open source, so you can also download it and build locally.

Minimal example

import numpy as np
import imageio
import invertedai as iai

location = "canada:vancouver:drake_street_and_pacific_blvd"  # select one of available locations

iai.add_apikey('')  # specify your key here or through the IAI_API_KEY variable

# get static information about a given location including map in osm
# format and list traffic lights with their IDs and locations.
location_info_response = iai.location_info(location=location)

# get traffic light states
light_response = iai.light(location=location)

# initialize the simulation by spawning NPCs
response = iai.initialize(
    location=location,  # select one of available locations
    agent_count=10,    # number of NPCs to spawn
    get_birdview=True,  # provides simple visualization - don't use in production
    traffic_light_state_history=[light_response.traffic_lights_states],  # provide traffic light states
)
agent_attributes = response.agent_attributes  # get dimension and other attributes of NPCs

images = [response.birdview.decode()]  # images storing visualizations of subsequent states
for _ in range(100):  # how many simulation steps to execute (10 steps is 1 second)

    # get next traffic light state
    light_response = iai.light(location=location, recurrent_states=light_response.recurrent_states)

    # query the API for subsequent NPC predictions
    response = iai.drive(
        location=location,
        agent_attributes=agent_attributes,
        agent_states=response.agent_states,
        recurrent_states=response.recurrent_states,
        get_birdview=True,
        traffic_lights_states=light_response.traffic_lights_states,
    )

    # save the visualization - requires np and cv2
    images.append(response.birdview.decode())

# save the visualization to disk
imageio.mimsave("iai-example.gif", np.array(images), format="GIF-PIL")

Stateful Cosimulation

Conceptually, the API is used to establish synchronous co-simulation between your own simulator running locally on your machine and the NPC engine running on Inverted AI servers. The basic integration in Python looks like this.

from typing import List
import numpy as np
import imageio
import invertedai as iai

# iai.add_apikey('')  # specify your key here or through the IAI_API_KEY variable


class LocalSimulator:
    """
    Mock up of a local simulator, where you control the ego vehicle. This example only supports single ego vehicle.
    """
    def __init__(self, ego_state: iai.AgentState, npc_states: List[iai.AgentState]):
        self.ego_state = ego_state
        self.npc_states = npc_states

    def _step_ego(self):
        """
        The simple motion model drives forward with constant speed.
        The ego agent ignores the map and NPCs for simplicity.
        """
        dt = 0.1
        dx = self.ego_state.speed * dt * np.cos(self.ego_state.orientation)
        dy = self.ego_state.speed * dt * np.sin(self.ego_state.orientation)

        self.ego_state = iai.AgentState(
            center=iai.Point(x=self.ego_state.center.x + dx, y=self.ego_state.center.y + dy),
            orientation=self.ego_state.orientation,
            speed=self.ego_state.speed,
        )

    def step(self, predicted_npc_states):
        self._step_ego()  # ego vehicle moves first so that it doesn't see future NPC movement
        self.npc_states = predicted_npc_states
        return self.ego_state


iai_simulation = iai.BasicCosimulation(  # instantiate a stateful wrapper for Inverted AI API
    location='canada:vancouver:ubc_roundabout',  # select one of available locations
    agent_count=5,  # how many vehicles in total to use in the simulation
    ego_agent_mask=[True, False, False, False, False],  # first vehicle is ego, rest are NPCs
    get_birdview=True,  # provides simple visualization - don't use in production
    traffic_lights=True,  # gets the traffic light states and used for initialization and steping the simulation
)
local_simulation = LocalSimulator(iai_simulation.ego_states[0], iai_simulation.npc_states)
images = [iai_simulation.birdview.decode()]  # images storing visualizations of subsequent states
for _ in range(100):  # how many simulation steps to execute (10 steps is 1 second)
    # query the API for subsequent NPC predictions, informing it how the ego vehicle acted
    iai_simulation.step([local_simulation.ego_state])
    # collect predictions for the next time step
    predicted_npc_behavior = iai_simulation.npc_states
    # execute predictions in your simulator, using your actions for the ego vehicle
    updated_ego_agent_state = local_simulation.step(predicted_npc_behavior)
    # save the visualization - requires np and cv2
    images.append(iai_simulation.birdview.decode())
# save the visualization to disk
imageio.mimsave("iai-example.gif", np.array(images), format="GIF-PIL")

To quickly check out how Inverted AI NPCs behave, try our Colab, where all agents are NPCs, or go to our github repository to execute it locally. When you're ready to try our NPCs with a real simulator, see the example CARLA integration. The examples are currently only provided in Python, but if you want to use the API from another language, you can use the REST API directly.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

invertedai-0.0.6.tar.gz (22.2 kB view hashes)

Uploaded Source

Built Distribution

invertedai-0.0.6-py3-none-any.whl (23.6 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page