Task Definitions¶

Comprehensive guide to the three-level task classification system for locomotion datasets.

Walking Tasks: Level • Incline • Decline
Stair Tasks: Ascent • Descent
Dynamic Tasks: Running • Jumping
Functional Tasks: Sit to Stand • Squats

Three-Level Task Classification System¶

Our data standard uses a hierarchical system to organize task information across three required columns:

task - Biomechanical category for grouping similar activities
task_id - Specific variant with primary distinguishing parameter
task_info - Detailed metadata in parseable key:value format

This system enables: - Efficient filtering by biomechanical category (task) - Quick selection of specific conditions (task_id) - Access to detailed experimental parameters (task_info)

Data Storage Format¶

Column Requirements¶

All phase-indexed and time-indexed datasets must include these columns:

Column	Type	Description	Example
`task`	string	Biomechanical category	`"incline_walking"`
`task_id`	string	Primary variant identifier	`"incline_10deg"`
`task_info`	string	Metadata in key:value format	`"incline_deg:10,speed_m_s:1.2"`

Task Info Format Specification¶

Format: "key1:value1,key2:value2,key3:value3"

Rules: - Keys contain units in the name (e.g., speed_m_s not speed) - Values are primitives only (float, int, boolean, string) - No spaces around colons or commas - Boolean values: true or false (lowercase) - Missing values: omit the key entirely

Type Conversion: - Numbers without decimals → integer - Numbers with decimals → float
- true/false → boolean - Everything else → string

Task Reference¶

Complete documentation for each standardized task type.

Quick Reference Table¶

Task	Task ID Examples	Primary Use
`level_walking`	`level`	Baseline gait analysis
`incline_walking`	`incline_5deg`, `incline_10deg`	Uphill locomotion
`decline_walking`	`decline_5deg`, `decline_10deg`	Downhill locomotion
`stair_ascent`	`stair_ascent`	Stair climbing analysis
`stair_descent`	`stair_descent`	Stair descent analysis
`run`	`run`	Running gait
`jump`	`jump`	Jumping mechanics
`sit_to_stand`	`sit_to_stand`	Functional mobility
`squats`	`squats`	Exercise analysis

Level Walking¶

Description: Walking on level ground (0° incline)

Task: level_walking
Task ID: level

Common Metadata:

Key	Type	Description	Typical Values
`speed_m_s`	float	Walking speed in m/s	0.8 - 1.8
`treadmill`	boolean	Treadmill vs overground	true/false
`overground`	boolean	Overground vs treadmill	true/false
`self_selected`	boolean	Self-selected pace	true/false

Example: "speed_m_s:1.2,treadmill:true"

Incline Walking¶

Description: Walking uphill on positive incline

Task: incline_walking
Task ID: incline_5deg, incline_10deg, incline_15deg, etc.

Common Metadata:

Key	Type	Description	Typical Values
`incline_deg`	float	Incline angle (positive)	5, 10, 15
`speed_m_s`	float	Walking speed in m/s	0.6 - 1.5
`treadmill`	boolean	Treadmill vs ramp	true/false

Example: "incline_deg:10,speed_m_s:1.0,treadmill:true"

Decline Walking¶

Description: Walking downhill on negative incline

Task: decline_walking
Task ID: decline_5deg, decline_10deg, decline_15deg, etc.

Common Metadata:

Key	Type	Description	Typical Values
`incline_deg`	float	Decline angle (negative)	-5, -10, -15
`speed_m_s`	float	Walking speed in m/s	0.6 - 1.5
`treadmill`	boolean	Treadmill vs ramp	true/false

Example: "incline_deg:-10,speed_m_s:0.8,treadmill:true"

Stair Ascent¶

Description: Walking up stairs

Task: stair_ascent
Task ID: stair_ascent

Common Metadata:

Key	Type	Description	Typical Values
`steps`	int	Number of steps	4, 8, 12
`height_m`	float	Step height in meters	0.15 - 0.20
`depth_m`	float	Step depth in meters	0.25 - 0.30
`speed_m_s`	float	Ascent speed if measured	0.4 - 0.8
`handrail`	boolean	Handrail availability/use	true/false
`step_over_step`	boolean	Reciprocal gait pattern	true/false

Example: "steps:8,height_m:0.17,handrail:false,step_over_step:true"

Stair Descent¶

Description: Walking down stairs

Task: stair_descent
Task ID: stair_descent

Common Metadata:

Key	Type	Description	Typical Values
`steps`	int	Number of steps	4, 8, 12
`height_m`	float	Step height in meters	0.15 - 0.20
`depth_m`	float	Step depth in meters	0.25 - 0.30
`speed_m_s`	float	Descent speed if measured	0.3 - 0.7
`handrail`	boolean	Handrail availability/use	true/false
`step_over_step`	boolean	Reciprocal gait pattern	true/false

Example: "steps:8,height_m:0.17,handrail:true,step_over_step:true"

Running¶

Description: Running or jogging locomotion

Task: run
Task ID: run

Common Metadata:

Key	Type	Description	Typical Values
`speed_m_s`	float	Running speed in m/s	2.0 - 5.0
`treadmill`	boolean	Treadmill vs overground	true/false
`incline_deg`	float	Incline if applicable	-5 to 15

Example: "speed_m_s:3.0,treadmill:true,incline_deg:0"

Jumping¶

Description: Various jumping activities

Task: jump
Task ID: jump

Common Metadata:

Key	Type	Description	Typical Values
`type`	string	Jump type	"vertical", "broad", "drop", "squat"
`height_m`	float	Jump or drop height	0.2 - 0.6
`distance_m`	float	Horizontal distance (broad jump)	1.0 - 2.5
`repetitions`	int	Number of jumps	1, 3, 5
`bilateral`	boolean	Two-leg vs single-leg	true/false

Example: "type:vertical,height_m:0.3,repetitions:1,bilateral:true"

Sit to Stand¶

Description: Rising from seated position

Task: sit_to_stand
Task ID: sit_to_stand

Common Metadata:

Key	Type	Description	Typical Values
`repetitions`	int	Number of repetitions	1, 5, 10
`chair_height_m`	float	Seat height in meters	0.40 - 0.50
`arms_used`	boolean	Whether arms were used	true/false
`speed`	string	Movement speed	"normal", "fast", "slow"

Example: "repetitions:5,chair_height_m:0.45,arms_used:false,speed:normal"

Squats¶

Description: Squatting exercise motion

Task: squats
Task ID: squats

Common Metadata:

Key	Type	Description	Typical Values
`repetitions`	int	Number of squats	5, 10, 15
`depth`	string	Squat depth	"partial", "parallel", "full"
`load_kg`	float	Additional weight if any	0 - 50
`tempo`	string	Movement tempo	"normal", "slow", "explosive"

Example: "repetitions:10,depth:parallel,load_kg:0,tempo:normal"

Usage Examples¶

Python Implementation¶

from user_libs.python.locomotion_data import LocomotionData

# Load dataset
data = LocomotionData('converted_datasets/dataset_phase.parquet')

# Parse task_info string into dictionary
info = data.parse_task_info("incline_deg:10,speed_m_s:1.2,treadmill:true")
# Returns: {'incline_deg': 10, 'speed_m_s': 1.2, 'treadmill': True}

# Filter by specific task_id
incline_10_data = data.filter_by_task_id('incline_10deg')

# Get specific metadata value for a subject-task combination
speed = data.get_task_metadata('SUB01', 'incline_walking', 'speed_m_s')
# Returns: 1.2

# Find all unique task_ids in dataset
task_ids = data.df['task_id'].unique()

MATLAB Implementation¶

% Load parquet file
data = parquetread('dataset_phase.parquet');

% Filter by task
incline_data = data(strcmp(data.task, 'incline_walking'), :);

% Parse task_info (simple approach)
task_info_str = incline_data.task_info{1};
% Parse "incline_deg:10,speed_m_s:1.2" into struct
pairs = strsplit(task_info_str, ',');
for i = 1:length(pairs)
    kv = strsplit(pairs{i}, ':');
    metadata.(kv{1}) = str2double(kv{2});
end

Implementation Notes¶

For Dataset Converters¶

Map native task names to our standard task categories
Extract primary parameter for task_id (e.g., incline angle)
Format metadata as comma-separated key:value pairs
Ensure consistency - same task should have same task_id format

For Data Users¶

Use task for broad filtering - get all walking data regardless of incline
Use task_id for specific conditions - get only 10° incline walking
Parse task_info for detailed analysis - extract exact speeds, dimensions
Handle missing metadata gracefully - not all datasets have all parameters

Standard Compliance¶

All three columns (task, task_id, task_info) are required
Use exact task names from this reference
Follow the key:value format strictly for parsing compatibility
Document any dataset-specific metadata keys

Future Extensions¶

Additional tasks can be added following this pattern: 1. Define the biomechanical category (task) 2. Identify the primary distinguishing parameter (task_id) 3. Document common metadata keys for task_info 4. Update converters to map to the new task

For questions about task classification or to propose new task types, see the Contributing Guide.

Task Definitions¶

Quick Navigation¶

Three-Level Task Classification System¶

Data Storage Format¶

Column Requirements¶

Task Info Format Specification¶

Task Reference¶

Quick Reference Table¶

Level Walking¶

Incline Walking¶

Decline Walking¶

Stair Ascent¶

Stair Descent¶

Running¶

Jumping¶

Sit to Stand¶

Squats¶

Usage Examples¶

Python Implementation¶

MATLAB Implementation¶

Implementation Notes¶

For Dataset Converters¶

For Data Users¶

Standard Compliance¶

Future Extensions¶