Adaickalavan
commented
2 years ago Examples on
- how to use SMARTS environment, and
- how to train agents in SMARTS environment using reinforcement learning (rl) algorithms
are given in the /SMARTS/examples folder.
For example, see
/SMARTS/examples/driving_in_traffic- https://github.com/huawei-noah/SMARTS/tree/example-driving-in-traffic/examples/driving_in_traffic
- Trains an agent to drive as fast and as far as possible, without colliding or going off-road, in heavy traffic using DreamerV2.
/SMARTS/examples/rllib- https://github.com/huawei-noah/SMARTS/tree/develop/examples/rllib
- Trains agents using RLlib library.
/SMARTS/examples/ray_multi_instance- https://github.com/huawei-noah/SMARTS/blob/develop/examples/ray_multi_instance.py
- Trains agents using (i) Ray to distribute training and evaluation, and (ii) PyTorch to implement the agent policy.
/SMARTS/baselines/marl_benchmark- https://github.com/huawei-noah/SMARTS/tree/develop/baselines/marl_benchmark
- This directory contains the scenarios, training environment, and agents used in the CoRL20 paper.
Additionally, to speed up data collection, you may consider using the provided ParallelEnv wrapper to simulate multiple SMARTS environments in parallel. Usage of ParallelEnv is illustrated in
/SMARTS/examples/parallel_environment.py- https://github.com/huawei-noah/SMARTS/blob/develop/examples/parallel_environment.py
We hope the above addresses your question and thus this issue is being closed. Please feel free to reopen this issue if your problem persists or open a new issue for other questions.
High Level Description [find the RL training section while running the code in example directory]
Desired SMARTS version [0.4.12]
Operating System [ubuntu 20.04]
Problems [I wanna design the RL training part of the simulation in example directory. However I cannot find it? Could anyone tell me which directory to look for?]