Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Deepdrive is a simulator that allows anyone with a PC to push the state-of-the-art in self-driving
| Date | Stars |
|---|---|
| 2026-07-24 | 924 |
| 2026-07-25 | 924 |
| 2026-07-28 | 924 |
| 2026-07-30 | 924 |
| 2026-08-06 | 924 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Deepdrive [](https://travis-ci.org/deepdrive/deepdrive) The easiest way to experiment with self-driving AI ## Simulator requirements - Linux - Python 3.6+ - 10GB disk space - 8GB RAM ## Optional - baseline agent requirements - CUDA capable GPU (tested and developed on 970, 1070, and 1060's) - 1.7 <= Tensorflow < 2.0 [See Tensorflow install tips](#tensorflow-install-tips) ## Install 1. [Create a Miniconda env](/docs/miniconda.md) 2. Clone ``` git clone https://github.com/deepdrive/deepdrive cd deepdrive ``` > Optional - Activate the Python conda env or virtualenv with Tensorflow installed, then > Note: If you use Anaconda (we recommend Miniconda) - only use `pip install` in your deepdrive conda environment, never `conda install` 3. Install ``` python install.py # Do not run as sudo! Use Miniconda or virtualenv to install without sudo. ``` #### Cloud [Cloud setup instructions](/docs/tutorial/cloud/cloud.md) ### Examples #### Forward-agent ``` python example.py ``` #### Synchronous forward-agent ``` python example_sync.py ``` * [Remote agent example](https://github.com/deepdrive/forward-agent) - operates over the network using the [deepdrive remote api](https://github.com/deepdrive/deepdrive-api) #### Mnet2 baseline agent ``` python main.py --mnet2-baseline --experiment my-baseline-test ``` #### Built-in C++ [FSM](https://github.com/deepdrive/deepdrive-sim/tree/c2d26a38692f1db61d48986263b20721ab136fe3/Plugins/DeepDrivePlugin/Source/DeepDrivePlugin/Private/Simulation/Agent/Controllers/LocalAI/States) / [PID](https://github.com/deepdrive/deepdrive-sim/blob/v3/Plugins/DeepDrivePlugin/Source/DeepDrivePlugin/Private/Simulation/Agent/Controllers/DeepDriveAgentSteeringController.cpp) agent that can overtake in the canyons map ``` python main.py --path-follower --experiment my-path-follower-test ``` #### Record training data for imitation learning / behavioral cloning ``` python main.py --record --jitter-actions --sync ``` Note that we recorded the baseline dataset in sync mode which is much slower than async mode. Async mode _probably_ is fine to record in, we just haven't got around to trying it out for v3. Optional: Convert to HDF5 files to tfrecords (for training MNET2) ``` python main.py --hdf5-2-tfrecord ``` #### Train on recorded data ``` python main.py --train [--agent dagger|dagger_mobilenet_v2|bootstrapped_ppo2] --recording-dir <your-hdf5-or-tfrecord-dir> ``` #### Train on our dataset Grab the [dataset](#dataset) ``` python main.py --train --recording-dir <the-directory-with-the-dataset> [--agent dagger|dagger_mobilenet_v2|bootstrapped_ppo2] ``` #### Tensorboard ``` tensorboard --logdir="<your-deepdrive-home>/tensorflow" ``` Where `<your-deepdrive-home>` below is by default in `$HOME/Deepdrive` and can be configured in `$HOME/.deepdrive/deepdrive_dir` #### Running unit tests ``` pytest tests/unit_tests/test_sanity.py ``` ### Key binds * <kbd>Esc</kbd> - Pause (Quit in Unreal Editor) * <kbd>Enter</kbd> - Pause with no menu * <kbd>P</kbd> - Pause in Unreal Editor * <kbd>1</kbd> - Chase cam * <kbd>2</kbd> - Orbit (side) cam * <kbd>3</kbd> - Hood cam * <kbd>4</kbd> - Free cam (use WASD to fly) * <kbd>Space</kbd> - Handbrake * <kbd>Alt+Tab</kbd> - Control other windows / Show mouse * <kbd>`</kbd> - Unreal console - do things like `stat FPS` * <kbd>M</kbd> - Drive the car with the keyboard WASD - be sure sync is off - Also known issue: Only works in path-follower mode right now * <kbd>Ctrl-number</kbd> - Change sun position - works for 1 => 7 * <kbd>B</kbd> - Show vehicle bounding boxes * <kbd>N</kbd> - Show vehicle collision boxes * <kbd>Page Up</kbd> - Next vehicle * <kbd>Page Down</kbd> - Prev vehicle ## Observation data All values returned in the observation keep Unreal conventions, specifically * All distances are in centimeters per Unreal's default data type * All rotations / angular values are in the order of
Excerpt of 12,806 characters
Read on GitHub
820
5
3
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:fd1723bf39a9d34e, topic:self-driving-car, topic:simulation, desc:self-driving
matched fp:fd1723bf39a9d34e, topic:deep-learning, topic:tensorflow
matched fp:fd1723bf39a9d34e, topic:reinforcement-learning, topic:deep-reinforcement-learning, readme:imitation learning