Dataset Viewer
Auto-converted to Parquet Duplicate
x
float64
0.1
0.9
y
float64
0.1
0.9
angle
int64
5
85
action
stringclasses
3 values
0.2
0.5
30
forward
0.1
0.7
80
left
0.9
0.1
10
right
0.4
0.6
45
forward
0.3
0.8
70
left
0.6
0.3
15
forward
0.8
0.2
5
right
0.5
0.5
60
left
0.7
0.4
25
forward
0.2
0.9
85
left

Humanoid Robot Navigation Dataset

Overview

This dataset contains state–action pairs created for humanoid robot navigation tasks.

It is designed for simple robotics learning scenarios such as:

  • Behavior cloning
  • Reinforcement learning
  • Policy learning
  • Navigation control experiments

The dataset maps robot state information (position and orientation) to discrete navigation actions.

Dataset Structure

File: train.csv

Each row represents one timestep observation and its corresponding action.

Columns

Column Type Description
x float Robot x position
y float Robot y position
angle int Heading angle in degrees
action string Navigation command (forward, left, right)

Example

x,y,angle,action
0.2,0.5,30,forward

Intended Use

This dataset can be used for:

  • Humanoid robot locomotion experiments
  • Behavior cloning baselines
  • Reinforcement learning tutorials
  • Robotics research demos
  • Educational purposes

License

MIT License

Author

Created for humanoid robotics training and experimentation.

Downloads last month
15