This project concerns how we may design environments in order to facilitate the training of deep reinforcement learning based autonomous driving agents. The goal of the project is to provide a working deep reinforcement learning framework that can learn to drive in visually complex environments, with a focus on providing a solution that:

1. Works out-of-the-box.
2. Learns in a short time to make it easier to quickly iterate on and test hypotheses.
3. Provide tailored metrics to compare agents between runs.

We have used the urban driving simulator CALRA (version 0.9.5) as our environment.

Find a detailed project write-up here.

Video of results:

Use the timestamps in the description to navigate to the experiments of your interest.

Figure 1: Town07 lap

• We provide two gym-like environments for CARLA*:
  1. Lap environment: This environment is focused on training an agent to follow a predetermined lap (see CarlaEnv/carla_lap_env.py)
  2. Route environment: This environment is focused on training agents that can navigate from point A to point B (see CarlaEnv/carla_route_env.py.
• We provide analysis of optimal PPO parameters, environment designs, reward functions, etc. with the aim of finding the optimal setup to train reinforcement learning based autonomous driving agents (see Chapter 4 of the project write-up for further details.)
• We have shown that how we train and use a VAE can be consequential to the performance of a deep reinforcement learning agent, and we have found that major improvements can be made by training the VAE to reconstruct semantic segmentation maps instead of reconstructing the RGB input itself.
• We have devised a model that can reliably solve the lap environment in ~8 hours on average on a Nvidia GTX 970.
• We have provided an example of how sub-policies can be used to navigate with PPO, and we found it to have moderate success in the route environment (See the sub-policy branch).

* While there are existing examples of gym-like environments for CARLA, there is no implementation that is officially endorsed by CARLA. Furthermore, most of the third-party environments do not provide an example of an agent that works out-of-the-box.

1. Learning to Drive in a Day by Kendall et. al. This paper by researchers at Wayve describes a method that showed how state representation learning through a variational autoencoder can be used to train a car to follow a straight country road in approximately 15 minutes.
2. Learning to Drive Smoothly in Minutes by Raffin et. al. This medium article lays out the details of a method that was able to train an agent in a Donkey Car simulator in only 5 minutes, using a similar approach as (1). They further provide some solutions to the unstable steering we may observe when we train with the straight forward speed-as-reward reward formulation of Kendall.
3. End-to-end Driving via Conditional Imitation Learning by Codevilla et. al. This paper outlines an imitation learning model that is able to learn to navigate arbitrary routes by using multiple actor networks, conditioned on what the current maneuver the vehicle should take is. We have used a similar approach in our route environment agent.

This is a high-level overview of the method.

1. Collect 10k 160x80x3 images by driving around manually.
2. Train a VAE to reconstruct the images.
3. Train an agent using the encoded state representations generated by the trained VAE and append a vector of measurements (steering, throttle, speed.) This is the input of the PPO-based agent.

Figure 2:

• Python 3.6
• CARLA 0.9.5 (may also work with later versions)
  • Our code expects the CARLA python API to be installed and available through import carla (see this)
  • We also recommend building a editor-less version of Carla by running the make package command in the root directory of CARLA.
  • Note that the map we use, Town07, may not be included by default when running make package. Add +MapsToCook=(FilePath="/Game/Carla/Maps/Town07") to Unreal/CarlaUE4/Config/DefaultGame.ini before running make package to solve this.
• TensorFlow for GPU (we have used version 1.13, may work with later versions)
• OpenAI gym (we used version 0.12.0)
• OpenCV for Python (we used version 4.0.0)
• A GPU with at least 4 GB VRAM (we used a GeForce GTX 970)

With the project, we provide a pretrained PPO agent for the lap environment. The checkpoint file for this model is located in the models folder.

The easiest way to get this model run, is to first set an environment variable named ${CARLA_ROOT} to point to the top-level directory in your CARLA installation.

Afterward, we can simply call:

python run_eval.py --model_name pretrained_agent -start_carla

And CARLA should automatically be started and our agent driving. This particular agent should be able to drive about 850m along the designated lap (Figure 1).

Note that our environment has only been designed to work with Town07 since this map is the one that closest resembles the environments of Kendall et. al. and Raffin et. al..

Set ${CARLA_ROOT} as is described in Running a Trained Agent.

Then use the following command to train a new agent:

python train.py --model_name name_of_your_model -start_carla

This will start training an agent with the default parameters, and checkpoint and log files will be written to models/name_of_your_model.

Recording of the evaluation episodes will also be written to models/name_of_your_model/videos by default, making it easier to evaluate an agent's behavior over time.

To view the training progress of an agent, and to compare trained agents in TensorBoard, use the following command:

tensorboard --logdir models/

If you wish to collect data to train the variational autoencoder yourself, you may use the following command:

python CarlaEnv/collect_data.py --output_dir vae/my_data -start_carla

Press SPACE to begin recording frames. 10K images will be saved by default.

After you have collected data to train the VAE with, use the following command to train the VAE:

cd vae
python train_vae.py --model_name my_trained_vae --dataset my_data

To view the training progress and to compare trained VAEs in TensorBoard, use the following command:

cd vae
tensorboard --logdir models/

Once we have a trained VAE, we can use the following command to inspect how its reconstructions look:

cd vae
python inspect_vae.py --model_dir models/my_trained_vae

Use the Set z by image button to seed your VAE with the latent z that is generated when the image selected is passed through the encoder (useful for comparing VAE reconstructions across models, as there is no guarantee that the features of the input will be encoded in the same indices of Z.)

We may also use the following command to see how a trained agent will behave to changes in latent space vector z by running:

python inspect_agent.py --model_name name_of_your_model

Here are a couple of ideas of how our work can be expanded or improved on:

• Temporal models such as World Models or other LSTM models
• Better state representations (higher resolution, sensor fusion with LiDAR, etc.)
• Improve exploration (random distillation networks, Ornstein-Uhlenbeck noise, etc.)
• Enforcing smooth driving e.g. through reward functions that penalize fluctuating actions
• Multi-agent training, or training with other vehicles on the road
• Making the agent obey traffic rules

• Seed does not make simulations deterministic, even in a synchronous environment
• Environment does not strictly confine to OpenAI gym's standard, meaning it cannot be used directly with their algorithms without modification

@mastersthesis{11250_2625841, title={Accelerating Training of Deep Reinforcement Learning-based Autonomous Driving Agents Through Comparative Study of Agent and Environment Designs}, url={https://ntnuopen.ntnu.no/ntnu-xmlui/handle/11250/2625841}, school={NTNU}, publisher={NTNU Open Access}, author={Vergara, Marcus Loo}, year={2019}, month={Oct}}