| Name | torchdrivesim JSON |
| Version |
0.2.1
JSON |
| download |
| home_page | None |
| Summary | TorchDriveSim is a lightweight 2D driving simulator, built entirely in PyTorch, primarily intended as a training environment for developing autonomous driving algorithms |
| upload_time | 2024-04-26 00:12:09 |
| maintainer | None |
| docs_url | None |
| author | None |
| requires_python | >=3.8 |
| license | Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for use, reproduction, and distribution as defined by Sections 1 through 9 of this document. "Licensor" shall mean the copyright owner or entity authorized by the copyright owner that is granting the License. "Legal Entity" shall mean the union of the acting entity and all other entities that control, are controlled by, or are under common control with that entity. For the purposes of this definition, "control" means (i) the power, direct or indirect, to cause the direction or management of such entity, whether by contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the outstanding shares, or (iii) beneficial ownership of such entity. "You" (or "Your") shall mean an individual or Legal Entity exercising permissions granted by this License. "Source" form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files. "Object" form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types. "Work" shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work (an example is provided in the Appendix below). "Derivative Works" shall mean any work, whether in Source or Object form, that is based on (or derived from) the Work and for which the editorial revisions, annotations, elaborations, or other modifications represent, as a whole, an original work of authorship. For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof. "Contribution" shall mean any work of authorship, including the original version of the Work and any modifications or additions to that Work or Derivative Works thereof, that is intentionally submitted to Licensor for inclusion in the Work by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, "submitted" means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Work, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as "Not a Contribution." "Contributor" shall mean Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by Licensor and subsequently incorporated within the Work. 2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute the Work and such Derivative Works in Source or Object form. 3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed. 4. Redistribution. You may reproduce and distribute copies of the Work or Derivative Works thereof in any medium, with or without modifications, and in Source or Object form, provided that You meet the following conditions: (a) You must give any other recipients of the Work or Derivative Works a copy of this License; and (b) You must cause any modified files to carry prominent notices stating that You changed the files; and (c) You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works; and (d) If the Work includes a "NOTICE" text file as part of its distribution, then any Derivative Works that You distribute must include a readable copy of the attribution notices contained within such NOTICE file, excluding those notices that do not pertain to any part of the Derivative Works, in at least one of the following places: within a NOTICE text file distributed as part of the Derivative Works; within the Source form or documentation, if provided along with the Derivative Works; or, within a display generated by the Derivative Works, if and wherever such third-party notices normally appear. The contents of the NOTICE file are for informational purposes only and do not modify the License. You may add Your own attribution notices within Derivative Works that You distribute, alongside or as an addendum to the NOTICE text from the Work, provided that such additional attribution notices cannot be construed as modifying the License. You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of Your modifications, or for any such Derivative Works as a whole, provided Your use, reproduction, and distribution of the Work otherwise complies with the conditions stated in this License. 5. Submission of Contributions. Unless You explicitly state otherwise, any Contribution intentionally submitted for inclusion in the Work by You to the Licensor shall be under the terms and conditions of this License, without any additional terms or conditions. Notwithstanding the above, nothing herein shall supersede or modify the terms of any separate license agreement you may have executed with Licensor regarding such Contributions. 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor, except as required for reasonable and customary use in describing the origin of the Work and reproducing the content of the NOTICE file. 7. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Work (and each Contributor provides its Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Work and assume any risks associated with Your exercise of permissions under this License. 8. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Work (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages. 9. Accepting Warranty or Additional Liability. While redistributing the Work or Derivative Works thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability. END OF TERMS AND CONDITIONS APPENDIX: How to apply the Apache License to your work. To apply the Apache License to your work, attach the following boilerplate notice, with the fields enclosed by brackets "[]" replaced with your own identifying information. (Don't include the brackets!) The text should be enclosed in the appropriate comment syntax for the file format. We also recommend that a file or class name and description of purpose be included on the same "printed page" as the copyright notice for easier identification within third-party archives. Copyright 2023 Inverted AI Ltd. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. |
| keywords |
torch
drive
sim
torch-drive-sim
torchdrivesim
invertedai
inverted ai
|
| VCS |
|
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
[pypi-badge]: https://badge.fury.io/py/torchdrivesim.svg
[pypi-link]: https://pypi.org/project/torchdrivesim/
[python-badge]: https://img.shields.io/pypi/pyversions/torchdrivesim.svg?color=%2334D058
[](https://github.com/inverted-ai/torchdrivesim/actions/workflows/CI_cpu.yml)
[![PyPI][pypi-badge]][pypi-link]
[![python-badge]][pypi-link]
[](https://docs.torchdrivesim.org/en/latest/)
# TorchDriveSim
<!-- start Features-->
TorchDriveSim is a lightweight 2D driving simulator, built entirely in [PyTorch](https://pytorch.org/), primarily intended as a training
environment for developing autonomous driving algorithms. Its main features are:
1. Fully differentiable execution producing a single computation graph, including state transition (kinematic models) and observation (differentiable rendering) models.
2. First class support for batch processing, down to the lowest level.
3. Support for heterogeneous agent types (vehicles, pedestrians, cyclists, etc.), each with its own kinematic model.
4. Extensible and customizable implementation of kinematic models (unconstrained, unicycle, bicycle, etc.), rendering modes, and rendering backends.
5. Support for extensible traffic control types, including traffic lights.
6. Differentiable implementations of various infraction metrics (collisions, off-road, wrong way).
7. Modular collection of wrappers modifying the simulator's behavior.
8. Ability to ingest any map in [Lanelet2](https://github.com/fzi-forschungszentrum-informatik/Lanelet2) format out of the box.
9. Integration with [IAI API](https://docs.inverted.ai/en/latest/) for initializing agent states and providing realistic behaviors.
<!-- end Features-->
### ** Warning **
The PyPI version of torchdrivesim (which is what you get when you run `pip install torchdrivesim`) comes equipped
only with the slower OpenCV renderer. Faster renderers are available, but require either
[`pytorch3d`](https://github.com/facebookresearch/pytorch3d/) or
[`nvdiffrast`](https://nvlabs.github.io/nvdiffrast/)
to be installed. Their CUDA dependencies can be tricky to satisfy, so we provide a suitable Dockerfile.
<!-- start readme-->
## Simulator Architecture
The simulated world consists of the following elements:
1. Static background, by default including road (drivable surface) and lane markings, represented as a triangular mesh.
2. Control elements, represented by rectangles with internal state. The simulator does not enforce their semantics.
3. A collection of agents grouped into arbitrary types. All agents are rigid rectangles.
4. Per agent type kinematic models, defining the agents' action space and how actions translate into motion.
5. A configurable renderer, displaying the world form bird's eye view (birdview), using a customizable color map.
Each agent is described by its static attributes (length, width, and others as needed by kinematic models),
dynamic state (x, y, orientation, speed), and a flag (present mask) indicating whether a given agent is currently alive.
At each time step, the agents perform actions, which in turn determine
their next state. The simulator allows the agents to overlap, which is identified as a collision but not prohibited,
and the states of different agents do not influence each other except through the agents' actions. The simulator can
operate either in homogeneous mode (all agents are the same type and their states and actions are tensors), or in
heterogeneous mode (there are multiple agent types and their states and actions are dictionaries mapping agent
types to tensors). To support both modes, most operations are applied as functors, which lift a function acting on
a single agent type into a function acting on all agent types. However, this behavior should be transparent to users
who do not modify the simulator code.
The base simulator requires actions for all agents and does not modify their presence masks. For convenience, we provide
various wrappers modifying the simulator's behavior, such as by controlling a subset of agents (by replay or pre-defined
ontrollers), removing agents that exit the designated area, monitoring infractions, recording video, and many others.
Unless specified otherwise, the wrappers can be combined in different orders to reach desired effects.
## Behavioral Models
The hardest driving scenarios are those that require interactions with other road users. When building simulated
environments it is crucial to ensure that the other agents behave realistically, but achieving that is not easy and
TorchDriveSim is flexible in terms of how those other agents are controlled. We provide a simple heuristic that achieves
minimally sensible driving, noting that in most cases it will be unsatisfactory and requiring additional extensions.
We also facilitate log replay and demonstrate how to use recorded actions from the INTERACTION dataset. However,
such replay is not reactive, often resulting in spurious collisions.
For maximum realism and reactivity, we recommend using our (Inverted AI) API for generating realistic behaviors,
which is integrated with TorchDriveSim. This is a paid offering that requires an API key, which you can obtain by
contacting us. For academics, we may be able to offer free API keys.
## Maps and Map Formats
Several CARLA maps (`carla_Town01`, `carla_Town02`, `carla_Town03`, `carla_Town04`, `carla_Town06`, `carla_Town07`, `carla_Town10HD`)
are included in `torchdrivesim` itself and can be loaded
by name. To include other maps, place the files in the format described below somewhere in a folder referenced by the
`TDS_RESOURCE_PATH` environment variable. Generally, a map is defined by a folder with the following structure:
```
MAPNAME/
metadata.json # custom metadata format
MAPNAME.osm # road network in Lanelet2 format
MAPNAME_mesh.json # custom road mesh format, can be derived from the .osm file
MAPNAME_stoplines.json # custom format specifying traffic lights, stop signs, and yield signs, if needed
```
See the bundled maps in `torchdrivesim/resources/maps` for examples. There is currently no tooling available
for creating TorchDriveSim-compatible maps, but you can try the experimental OpenDRIVE
[converter](https://github.com/inverted-ai/map-converter).
## Scenario Definition
With maps and behavioral models available, the final hurdle is to define a suite of driving scenarios that can be
used for testing and evaluation. TorchDriveSim provides helpers for initializing the simulation state, including
by calling Inverted AI API, instantiating from a log (when available), and using some simple heuristics. It also
provides functions for identifying driving infractions, specifically collisions, going off-road, and driving wrong way.
However, it does not specify goals or rewards, leaving that to the user. Over time, we are planning to release
various scenario suites that can serve as benchmarks.
## Kinematic models
The primary kinematic model for vehicles in TorchDriveSim is the bicycle model, where the action consists of steering
and acceleration. It requires additionally specifying the rear axis offset to control the vehicle turn radius, but
it does not use the front axis offset, since that can not be fit by observing the vehicle movement from the outside,
effectively assuming the front axis is in the middle of the vehicle. Other kinematic models available are the
unconstrained model, where the action is the state delta between subsequent time steps, and the teleporting model,
where the action directly specifies the next state. We also provide different variations of those models, and it is
straightforward to implement a custom one.
## Differentiable rendering
TorchDriveSim supports three rendering backends, using cv2, pytorch3d, and nvdiffrast, respectively. The images produced
by all three look very similar and differ only in the details of how different triangles are pixelated.
The cv2 backend is the easiest to install and it is included as a required dependency. Pytorch3d and nvdiffrast
need to be installed separately, as per the instructions below. We also provide a dummy rendering backend that
returns an empty image, mostly for debugging and benchmarking purposes.
## Installation
Running `pip install torchdrivesim` only provides access to the basic OpenCV renderer. To be able to use the faster
pytorch3d renderer, make sure to first install the correct versions of `torch` and `pytorch3d` using the instructions
below. You can also install [`nvdiffrast`](https://nvlabs.github.io/nvdiffrast/#installation), which can be even faster,
but it is also subject to more restrictive license conditions.
Generally, the more images you render in parallel (either by having more ego agents or larger batches), the more
of a gain you will get from the faster renderers.
To install the correct `torch` using `pip`, go visit the
[prebuilt whls page](https://download.pytorch.org/whl/torch_stable.html) to select the right `.whl` file based on the
Python version, cuda availability and the operating system. For example, to install version `1.11.0` for `python3.8`
with `cuda` on Linux, run
`pip install https://download.pytorch.org/whl/cu113/torch-1.10.2%2Bcu113-cp38-cp38-linux_x86_64.whl`
or
`pip install torch==1.11.0+cu113 -f https://download.pytorch.org/whl/torch_stable.html`
To install the correct `pytorch3d`, the user need to find the correct prebuilt wheel for the installed `torch` version,
more details can be found at the
[official installation page](https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md). For example, to
install version `0.7.2` for `python3.8` with `pip`, run
`pip install https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/pytorch3d-0.7.2-cp38-cp38-linux_x86_64.whl`
or
`pip install pytorch3d==0.7.2 -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/download.html`
Here are the summarized example steps for python3.8 and cuda 11.3:
```bazaar
1. pip install torch==1.11.0+cu113 -f https://download.pytorch.org/whl/torch_stable.html
2. pip install pytorch3d==0.7.2 -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/download.html
3. pip install torchdrivesim
```
## Docker
In order to use TorchDriveSim smoothly without worrying about installing dependencies, we provide a
[Dockerfile](Dockerfile) that works either with or without gpu. In order to build the docker image,
run `docker build --target torchdrivesim -t torchdrivesim:latest . `. To run the container with GPU access,
run `docker run --runtime=nvidia -it torchdrivesim:latest /bin/bash`. To run the container without GPU access,
run `docker run -it torchdrivesim:latest /bin/bash`. For more information regarding setting up GPU runtime with Docker,
follow [this official link](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html)
for the installation process.
## Citations
If you use TorchDriveSim in your research, please cite the following [paper](https://arxiv.org/abs/2104.11212),
for which an early version of TorchDriveSim was initially developed.
```bibtex
@INPROCEEDINGS{itra2021,
author={\'Scibior, Adam and Lioutas, Vasileios and Reda, Daniele and Bateni, Peyman and Wood, Frank},
booktitle={2021 IEEE International Intelligent Transportation Systems Conference (ITSC)},
title={Imagining The Road Ahead: Multi-Agent Trajectory Prediction via Differentiable Simulation},
year={2021},
pages={720-725},
doi={10.1109/ITSC48978.2021.9565113}}
```
<!-- end readme-->
Raw data
{
"_id": null,
"home_page": null,
"name": "torchdrivesim",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "torch, drive, sim, torch-drive-sim, torchdrivesim, invertedai, inverted ai",
"author": null,
"author_email": "\"Inverted AI Ltd.\" <info@inverted.ai>",
"download_url": "https://files.pythonhosted.org/packages/ae/68/8b15340f46a2b23ae95605f11b7fd8a86136fce86de55159db7fcee3c18e/torchdrivesim-0.2.1.tar.gz",
"platform": null,
"description": "[pypi-badge]: https://badge.fury.io/py/torchdrivesim.svg\n[pypi-link]: https://pypi.org/project/torchdrivesim/ \n[python-badge]: https://img.shields.io/pypi/pyversions/torchdrivesim.svg?color=%2334D058\n[](https://github.com/inverted-ai/torchdrivesim/actions/workflows/CI_cpu.yml)\n[![PyPI][pypi-badge]][pypi-link]\n[![python-badge]][pypi-link]\n[](https://docs.torchdrivesim.org/en/latest/)\n\n# TorchDriveSim\n<!-- start Features-->\nTorchDriveSim is a lightweight 2D driving simulator, built entirely in [PyTorch](https://pytorch.org/), primarily intended as a training\nenvironment for developing autonomous driving algorithms. Its main features are:\n1. Fully differentiable execution producing a single computation graph, including state transition (kinematic models) and observation (differentiable rendering) models.\n2. First class support for batch processing, down to the lowest level.\n3. Support for heterogeneous agent types (vehicles, pedestrians, cyclists, etc.), each with its own kinematic model.\n4. Extensible and customizable implementation of kinematic models (unconstrained, unicycle, bicycle, etc.), rendering modes, and rendering backends.\n5. Support for extensible traffic control types, including traffic lights.\n6. Differentiable implementations of various infraction metrics (collisions, off-road, wrong way).\n7. Modular collection of wrappers modifying the simulator's behavior.\n8. Ability to ingest any map in [Lanelet2](https://github.com/fzi-forschungszentrum-informatik/Lanelet2) format out of the box.\n9. Integration with [IAI API](https://docs.inverted.ai/en/latest/) for initializing agent states and providing realistic behaviors.\n<!-- end Features-->\n### ** Warning **\nThe PyPI version of torchdrivesim (which is what you get when you run `pip install torchdrivesim`) comes equipped\nonly with the slower OpenCV renderer. Faster renderers are available, but require either \n[`pytorch3d`](https://github.com/facebookresearch/pytorch3d/) or \n[`nvdiffrast`](https://nvlabs.github.io/nvdiffrast/)\nto be installed. Their CUDA dependencies can be tricky to satisfy, so we provide a suitable Dockerfile.\n\n<!-- start readme-->\n## Simulator Architecture\nThe simulated world consists of the following elements:\n1. Static background, by default including road (drivable surface) and lane markings, represented as a triangular mesh.\n2. Control elements, represented by rectangles with internal state. The simulator does not enforce their semantics.\n3. A collection of agents grouped into arbitrary types. All agents are rigid rectangles.\n4. Per agent type kinematic models, defining the agents' action space and how actions translate into motion.\n5. A configurable renderer, displaying the world form bird's eye view (birdview), using a customizable color map.\n\nEach agent is described by its static attributes (length, width, and others as needed by kinematic models),\ndynamic state (x, y, orientation, speed), and a flag (present mask) indicating whether a given agent is currently alive.\nAt each time step, the agents perform actions, which in turn determine\ntheir next state. The simulator allows the agents to overlap, which is identified as a collision but not prohibited,\nand the states of different agents do not influence each other except through the agents' actions. The simulator can\noperate either in homogeneous mode (all agents are the same type and their states and actions are tensors), or in\nheterogeneous mode (there are multiple agent types and their states and actions are dictionaries mapping agent\ntypes to tensors). To support both modes, most operations are applied as functors, which lift a function acting on\na single agent type into a function acting on all agent types. However, this behavior should be transparent to users\nwho do not modify the simulator code.\n\nThe base simulator requires actions for all agents and does not modify their presence masks. For convenience, we provide\nvarious wrappers modifying the simulator's behavior, such as by controlling a subset of agents (by replay or pre-defined\nontrollers), removing agents that exit the designated area, monitoring infractions, recording video, and many others.\nUnless specified otherwise, the wrappers can be combined in different orders to reach desired effects.\n\n## Behavioral Models\n\nThe hardest driving scenarios are those that require interactions with other road users. When building simulated\nenvironments it is crucial to ensure that the other agents behave realistically, but achieving that is not easy and\nTorchDriveSim is flexible in terms of how those other agents are controlled. We provide a simple heuristic that achieves\nminimally sensible driving, noting that in most cases it will be unsatisfactory and requiring additional extensions.\nWe also facilitate log replay and demonstrate how to use recorded actions from the INTERACTION dataset. However,\nsuch replay is not reactive, often resulting in spurious collisions.\n\nFor maximum realism and reactivity, we recommend using our (Inverted AI) API for generating realistic behaviors,\nwhich is integrated with TorchDriveSim. This is a paid offering that requires an API key, which you can obtain by\ncontacting us. For academics, we may be able to offer free API keys.\n\n## Maps and Map Formats\n\nSeveral CARLA maps (`carla_Town01`, `carla_Town02`, `carla_Town03`, `carla_Town04`, `carla_Town06`, `carla_Town07`, `carla_Town10HD`)\nare included in `torchdrivesim` itself and can be loaded\nby name. To include other maps, place the files in the format described below somewhere in a folder referenced by the\n`TDS_RESOURCE_PATH` environment variable. Generally, a map is defined by a folder with the following structure:\n```\nMAPNAME/\n metadata.json # custom metadata format\n MAPNAME.osm # road network in Lanelet2 format\n MAPNAME_mesh.json # custom road mesh format, can be derived from the .osm file\n MAPNAME_stoplines.json # custom format specifying traffic lights, stop signs, and yield signs, if needed\n```\n\nSee the bundled maps in `torchdrivesim/resources/maps` for examples. There is currently no tooling available\nfor creating TorchDriveSim-compatible maps, but you can try the experimental OpenDRIVE\n[converter](https://github.com/inverted-ai/map-converter).\n\n## Scenario Definition\n\nWith maps and behavioral models available, the final hurdle is to define a suite of driving scenarios that can be\nused for testing and evaluation. TorchDriveSim provides helpers for initializing the simulation state, including\nby calling Inverted AI API, instantiating from a log (when available), and using some simple heuristics. It also\nprovides functions for identifying driving infractions, specifically collisions, going off-road, and driving wrong way.\nHowever, it does not specify goals or rewards, leaving that to the user. Over time, we are planning to release\nvarious scenario suites that can serve as benchmarks.\n\n## Kinematic models\n\nThe primary kinematic model for vehicles in TorchDriveSim is the bicycle model, where the action consists of steering\nand acceleration. It requires additionally specifying the rear axis offset to control the vehicle turn radius, but\nit does not use the front axis offset, since that can not be fit by observing the vehicle movement from the outside,\neffectively assuming the front axis is in the middle of the vehicle. Other kinematic models available are the\nunconstrained model, where the action is the state delta between subsequent time steps, and the teleporting model,\nwhere the action directly specifies the next state. We also provide different variations of those models, and it is\nstraightforward to implement a custom one.\n\n## Differentiable rendering\n\nTorchDriveSim supports three rendering backends, using cv2, pytorch3d, and nvdiffrast, respectively. The images produced\nby all three look very similar and differ only in the details of how different triangles are pixelated.\nThe cv2 backend is the easiest to install and it is included as a required dependency. Pytorch3d and nvdiffrast\nneed to be installed separately, as per the instructions below. We also provide a dummy rendering backend that\nreturns an empty image, mostly for debugging and benchmarking purposes.\n\n## Installation\n\nRunning `pip install torchdrivesim` only provides access to the basic OpenCV renderer. To be able to use the faster\npytorch3d renderer, make sure to first install the correct versions of `torch` and `pytorch3d` using the instructions\nbelow. You can also install [`nvdiffrast`](https://nvlabs.github.io/nvdiffrast/#installation), which can be even faster,\nbut it is also subject to more restrictive license conditions.\nGenerally, the more images you render in parallel (either by having more ego agents or larger batches), the more\nof a gain you will get from the faster renderers.\n\nTo install the correct `torch` using `pip`, go visit the\n[prebuilt whls page](https://download.pytorch.org/whl/torch_stable.html) to select the right `.whl` file based on the\nPython version, cuda availability and the operating system. For example, to install version `1.11.0` for `python3.8`\nwith `cuda` on Linux, run \n `pip install https://download.pytorch.org/whl/cu113/torch-1.10.2%2Bcu113-cp38-cp38-linux_x86_64.whl` \nor \n`pip install torch==1.11.0+cu113 -f https://download.pytorch.org/whl/torch_stable.html` \n\nTo install the correct `pytorch3d`, the user need to find the correct prebuilt wheel for the installed `torch` version,\nmore details can be found at the\n[official installation page](https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md). For example, to \ninstall version `0.7.2` for `python3.8` with `pip`, run \n`pip install https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/pytorch3d-0.7.2-cp38-cp38-linux_x86_64.whl` \nor \n`pip install pytorch3d==0.7.2 -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/download.html` \n\nHere are the summarized example steps for python3.8 and cuda 11.3:\n```bazaar\n1. pip install torch==1.11.0+cu113 -f https://download.pytorch.org/whl/torch_stable.html\n2. pip install pytorch3d==0.7.2 -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu113_pyt1110/download.html\n3. pip install torchdrivesim\n```\n## Docker\n\nIn order to use TorchDriveSim smoothly without worrying about installing dependencies, we provide a \n[Dockerfile](Dockerfile) that works either with or without gpu. In order to build the docker image,\nrun `docker build --target torchdrivesim -t torchdrivesim:latest . `. To run the container with GPU access,\nrun `docker run --runtime=nvidia -it torchdrivesim:latest /bin/bash`. To run the container without GPU access,\nrun `docker run -it torchdrivesim:latest /bin/bash`. For more information regarding setting up GPU runtime with Docker,\nfollow [this official link](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html)\nfor the installation process.\n\n## Citations\n\nIf you use TorchDriveSim in your research, please cite the following [paper](https://arxiv.org/abs/2104.11212),\nfor which an early version of TorchDriveSim was initially developed.\n\n```bibtex\n@INPROCEEDINGS{itra2021,\n author={\\'Scibior, Adam and Lioutas, Vasileios and Reda, Daniele and Bateni, Peyman and Wood, Frank},\n booktitle={2021 IEEE International Intelligent Transportation Systems Conference (ITSC)}, \n title={Imagining The Road Ahead: Multi-Agent Trajectory Prediction via Differentiable Simulation}, \n year={2021},\n pages={720-725},\n doi={10.1109/ITSC48978.2021.9565113}}\n```\n<!-- end readme-->\n",
"bugtrack_url": null,
"license": "Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. \"License\" shall mean the terms and conditions for use, reproduction, and distribution as defined by Sections 1 through 9 of this document. \"Licensor\" shall mean the copyright owner or entity authorized by the copyright owner that is granting the License. \"Legal Entity\" shall mean the union of the acting entity and all other entities that control, are controlled by, or are under common control with that entity. For the purposes of this definition, \"control\" means (i) the power, direct or indirect, to cause the direction or management of such entity, whether by contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the outstanding shares, or (iii) beneficial ownership of such entity. \"You\" (or \"Your\") shall mean an individual or Legal Entity exercising permissions granted by this License. \"Source\" form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files. \"Object\" form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types. \"Work\" shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work (an example is provided in the Appendix below). \"Derivative Works\" shall mean any work, whether in Source or Object form, that is based on (or derived from) the Work and for which the editorial revisions, annotations, elaborations, or other modifications represent, as a whole, an original work of authorship. For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof. \"Contribution\" shall mean any work of authorship, including the original version of the Work and any modifications or additions to that Work or Derivative Works thereof, that is intentionally submitted to Licensor for inclusion in the Work by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, \"submitted\" means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Work, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as \"Not a Contribution.\" \"Contributor\" shall mean Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by Licensor and subsequently incorporated within the Work. 2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute the Work and such Derivative Works in Source or Object form. 3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed. 4. Redistribution. You may reproduce and distribute copies of the Work or Derivative Works thereof in any medium, with or without modifications, and in Source or Object form, provided that You meet the following conditions: (a) You must give any other recipients of the Work or Derivative Works a copy of this License; and (b) You must cause any modified files to carry prominent notices stating that You changed the files; and (c) You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works; and (d) If the Work includes a \"NOTICE\" text file as part of its distribution, then any Derivative Works that You distribute must include a readable copy of the attribution notices contained within such NOTICE file, excluding those notices that do not pertain to any part of the Derivative Works, in at least one of the following places: within a NOTICE text file distributed as part of the Derivative Works; within the Source form or documentation, if provided along with the Derivative Works; or, within a display generated by the Derivative Works, if and wherever such third-party notices normally appear. The contents of the NOTICE file are for informational purposes only and do not modify the License. You may add Your own attribution notices within Derivative Works that You distribute, alongside or as an addendum to the NOTICE text from the Work, provided that such additional attribution notices cannot be construed as modifying the License. You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of Your modifications, or for any such Derivative Works as a whole, provided Your use, reproduction, and distribution of the Work otherwise complies with the conditions stated in this License. 5. Submission of Contributions. Unless You explicitly state otherwise, any Contribution intentionally submitted for inclusion in the Work by You to the Licensor shall be under the terms and conditions of this License, without any additional terms or conditions. Notwithstanding the above, nothing herein shall supersede or modify the terms of any separate license agreement you may have executed with Licensor regarding such Contributions. 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor, except as required for reasonable and customary use in describing the origin of the Work and reproducing the content of the NOTICE file. 7. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Work (and each Contributor provides its Contributions) on an \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Work and assume any risks associated with Your exercise of permissions under this License. 8. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Work (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages. 9. Accepting Warranty or Additional Liability. While redistributing the Work or Derivative Works thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability. END OF TERMS AND CONDITIONS APPENDIX: How to apply the Apache License to your work. To apply the Apache License to your work, attach the following boilerplate notice, with the fields enclosed by brackets \"[]\" replaced with your own identifying information. (Don't include the brackets!) The text should be enclosed in the appropriate comment syntax for the file format. We also recommend that a file or class name and description of purpose be included on the same \"printed page\" as the copyright notice for easier identification within third-party archives. Copyright 2023 Inverted AI Ltd. Licensed under the Apache License, Version 2.0 (the \"License\"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an \"AS IS\" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.",
"summary": "TorchDriveSim is a lightweight 2D driving simulator, built entirely in PyTorch, primarily intended as a training environment for developing autonomous driving algorithms",
"version": "0.2.1",
"project_urls": null,
"split_keywords": [
"torch",
" drive",
" sim",
" torch-drive-sim",
" torchdrivesim",
" invertedai",
" inverted ai"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "87cb248e1a64b8edb8571554a45d23cc5dbbbf80a2ad6e5840dbc2506a9be54c",
"md5": "89e6d3480f2c6dbee46f8c4917c13b57",
"sha256": "553d93fb96ad03c318312a3726c3443bb9cea14b6306b3eeb9a18a7c4988aaea"
},
"downloads": -1,
"filename": "torchdrivesim-0.2.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "89e6d3480f2c6dbee46f8c4917c13b57",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 23489637,
"upload_time": "2024-04-26T00:12:06",
"upload_time_iso_8601": "2024-04-26T00:12:06.229780Z",
"url": "https://files.pythonhosted.org/packages/87/cb/248e1a64b8edb8571554a45d23cc5dbbbf80a2ad6e5840dbc2506a9be54c/torchdrivesim-0.2.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "ae688b15340f46a2b23ae95605f11b7fd8a86136fce86de55159db7fcee3c18e",
"md5": "13e035b24c8241098d3bfdcd6e23d911",
"sha256": "ffb33118daf2b7d9dd814cade7fedcc89799ecf6fca4a2d4d9d488d3184d75d7"
},
"downloads": -1,
"filename": "torchdrivesim-0.2.1.tar.gz",
"has_sig": false,
"md5_digest": "13e035b24c8241098d3bfdcd6e23d911",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 51183489,
"upload_time": "2024-04-26T00:12:09",
"upload_time_iso_8601": "2024-04-26T00:12:09.432477Z",
"url": "https://files.pythonhosted.org/packages/ae/68/8b15340f46a2b23ae95605f11b7fd8a86136fce86de55159db7fcee3c18e/torchdrivesim-0.2.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-04-26 00:12:09",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "torchdrivesim"
}
