Implementation:Farama Foundation Gymnasium Walker2dEnv V4
| Knowledge Sources | |
|---|---|
| Domains | Reinforcement_Learning, MuJoCo_Environments |
| Last Updated | 2026-02-15 03:00 GMT |
Overview
Concrete implementation of the Walker2d v4 MuJoCo locomotion environment provided by Gymnasium.
Description
The Walker2d v4 environment implements the 2D bipedal walking robot using MuJoCo bindings (mujoco >= 2.1.3). The walker has seven body parts: torso, two thighs, two legs, and two feet, connected by six hinge joints. The goal is to walk forward by applying torques to the joints. The reward combines forward velocity, a healthy survival reward, and a control cost penalty. The episode terminates if the walker becomes unhealthy (z-coordinate or angle outside healthy ranges). Note that v4 has a known issue where the left and right feet have different friction values (0.9 and 1.9) and healthy_reward is given on every step even when unhealthy.
Usage
Use this environment for reproducing results from papers that used Walker2d-v4. For new research, consider Walker2d-v5 which fixes the foot friction asymmetry, the healthy_reward bug, and provides more detailed info dictionaries.
Code Reference
Source Location
- Repository: Farama_Foundation_Gymnasium
- File: gymnasium/envs/mujoco/walker2d_v4.py
Signature
class Walker2dEnv(MujocoEnv, utils.EzPickle):
def __init__(
self,
forward_reward_weight=1.0,
ctrl_cost_weight=1e-3,
healthy_reward=1.0,
terminate_when_unhealthy=True,
healthy_z_range=(0.8, 2.0),
healthy_angle_range=(-1.0, 1.0),
reset_noise_scale=5e-3,
exclude_current_positions_from_observation=True,
**kwargs,
)
Import
import gymnasium as gym
env = gym.make("Walker2d-v4")
I/O Contract
Inputs
| Name | Type | Required | Description |
|---|---|---|---|
| action | np.ndarray (6,) | Yes | Torques applied to thigh, leg, foot, left thigh, left leg, and left foot joints, range [-1, 1] |
Outputs
| Name | Type | Description |
|---|---|---|
| observation | np.ndarray (17,) | State vector: qpos (8 elements, x excluded), qvel (9 elements, clipped to [-10, 10]) |
| reward | float | forward_reward + healthy_reward - ctrl_cost |
| terminated | bool | True if walker is unhealthy (z or angle outside healthy ranges) |
| truncated | bool | Episode truncation (handled by TimeLimit wrapper) |
| info | dict | Contains x_position, x_velocity |
Usage Examples
import gymnasium as gym
env = gym.make("Walker2d-v4")
observation, info = env.reset(seed=42)
for _ in range(1000):
action = env.action_space.sample()
observation, reward, terminated, truncated, info = env.step(action)
if terminated or truncated:
observation, info = env.reset()
env.close()