webxos/BCI-grid-movement-v1

It was generated using a propietary application "webXOS GYM-1" found in the /gym/ folder. Use it create similar data sets. The app is made to run in one html page to generate similar datasets to the one listed here.

BCI Grid Movement Intent Dataset

Description

BCI dataset with WASD movement intent data from single-enemy grid training environment. Includes neural signals, movement vectors, and game telemetry for ML training on movement intent prediction.

Session Information

• Session ID: bcigrid_1767251898029
• Duration: 1m 3s
• Level Reached: 3
• Accuracy: 16.88%
• Avg Reaction Time: 3712.8ms
• Distance Traveled: 343.8m
• Movement Intent Samples: 1,355
• Neural Samples: 5,143

Environment

• Grid Size: 50x50 units
• Single Enemy System: One enemy at a time, respawns after destruction
• WASD Movement: Full 3D movement with intent tracking
• Fixed Player Height: 1.7 units

Data Structure - UNIFIED SCHEMA

Features (all rows have same columns):

• timestamp: UNIX timestamp in milliseconds
• session_time: Time since session start in ms
• data_type: Type of data: neural/movement/telemetry/event
• neural_channels: 12-channel simulated neural signals (only for data_type='neural')
• game_context: Game state context
• movement_intent: WASD key states (W/A/S/D booleans)
• fire_intent: Mouse click intent (0/1)
• player_position: [x, y, z] coordinates
• player_velocity: Velocity magnitude
• distance_traveled: Total distance traveled in meters
• movement_vector: [x, y, z] movement vector
• keys: Object with W/A/S/D key states
• game_state: Level, XP, score, enemies destroyed, current streak
• input_state: Input state data
• event_type: For events: game_start/enemy_spawn/shot_fired/enemy_destroyed/level_up
• event_data: Additional event-specific data

Intended Use

• Train ML models for movement intent prediction from neural signals
• Analyze correlations between WASD inputs and neural patterns
• Research reinforcement learning for BCI navigation
• Study 3D environment interaction in BCI systems

Loading Instructions

Hugging Face Format:

from datasets import load_dataset

# Load the dataset
dataset = load_dataset('json', data_files={
    'train': 'train/dataset.jsonl',
    'test': 'test/dataset.jsonl'
})

# Filter by data_type if needed
neural_data = dataset.filter(lambda x: x['data_type'] == 'neural')

Direct JSONL:

import json

with open('dataset.jsonl', 'r') as f:
    data = [json.loads(line) for line in f]

Tags

bci, movement-intent, neural-signals, fps-telemetry, grid-environment, brain-computer-interface, wasd-movement, intent-prediction

Citation

@misc{bcigrid_2026,
  title={BCI Grid Movement Intent Dataset},
  author={BCI Research},
  year=2026,
  note={Movement intent data from grid training environment}
}

License

MIT