What you get

• The Gymnasium API. reset, step and a seed; your training code does not need to know it is driving a simulator.
• Deterministic seeds. The same seed and the same actions give byte-identical observations.
• Reactive traffic. Other road users yield, avoid and obey signals, so a hard brake or a late merge has consequences.
• A PPO baseline. Trains in 38 minutes on one CPU with no GPU; a same-seed rerun matched its returns byte for byte.
• Sensor-only mode. The policy gets rendered camera passes and the clock, and no engine state.

Quickstart

State observations need no GPU. Sign in, install the package, the runtime and a map:

simforge login
pip install 'simforge-rl[ppo]'
simforge runtime install --accept-license
simforge maps pull richmond-field-station@v9 --accept-license

Then step eight episodes at once from Python:

from simforge_rl import RuntimeSessions, ScenarioSet, make_vector_env

tasks = ScenarioSet.bundled("richmond-field-station")
with RuntimeSessions(tasks, "--no-sensors", workdir="work") as sessions:
    envs = make_vector_env(tasks, sessions.transports(), num_envs=8)
    obs, info = envs.reset(seed=0)
    obs, reward, terminated, truncated, info = envs.step(envs.action_space.sample())

Actions, observations, rewards, task sets and camera mode are in the RL reference.

The PPO baseline

PPO on the 24 Richmond Field Station tasks, 1.23 M environment steps on one CPU. Mean held-out return rose from 0.79 to 2.74, against 0.71 for the scenarios' own authored driving. It wins on 12 of the 24 tasks and still crashes on some: