## Performance Metric

The purpose of a performance metric is to determine the effectiveness of various gaits, trajectories and foot designs in an concrete and objective manner. The Stanford force plate can measure 3 forces and 3 moments (6-axis) at a fixed sample rate (currently 1kHz), so any combinations of these measurements and time history can be used to extract performance information from a data set. This is a first pass at establishing some metrics to analyze test data beyond visual inspection.

Metric #1 (PM1) - This metric values net adhesive force (normal), only positive climbing forces (longitudinal) and positive (inward, toward body center for the current leg, which is set up as a right leg) lateral forces
Metric #2 (PM2) - This metric values net adhesive force (normal), net climbing forces (longitudinal) and inward lateral forces
Metric #3 (PM3) - This metric values low detachment forces (), net climbing forces (longitudinal) and inward lateral forces
Metric #4 (PM4) - This metric values net adhesive force (normal), net climbing forces (longitudinal), inward lateral forces and low disengagement forces ()

- Detachment force = net adhesive (normal) force over last 0.3s of climbing window
- Disengagement force = all residual forces 0.3s after end of climbing window

The XYZ axes for the force plate are as follows:

• +X is force to the left (for a right-side leg, this is pulling laterally toward the body center)
• +Y is force down (this is a climbing force, acting in the direction of gravity)
• +Z is pulling away from the wall (this is an adhesive force in the direction normal to the wall)
```  +X <------*
/|
/ |
/  |
/   |
+Z '    V +Y
```

Gait 13 - Decent gait using the dactyl on the platform
Gait 28F - Early gait using pulling in (wing angle) on the platform (front leg trajectory)
Gait 46F - Decent gait using the v0.1? stanford feet on the platform (front leg trajectory)
Gait 47 - Decent gait using the v0.1 stanford feet on the climbing wall

### Summary #1 - 8/11/04 Testing

On this day, 4 different gaits were compared for 2 foot configurations:
1. the unchanged foot
2. a foot with a yaw stop and 1 bent claw

Notes
• each group contains 3 numbered tests with identical settings (ie. 1,2,3 for no changes, 4,5,6 for yaw stop & bent claw)
• shows how gaits are getting progressively better (MAIN POINT HERE)
• concrete measure of each gait/foot configuration performance
• larger performance metric numbers mean better results
• shows the simpler (only 2 cases for each) metric between 4 standard gaits
• shows the difference between no foot changes and yaw stop & 1 bent claw
• generally grouped together
• performance metrics are NOT doing a good job of capturing detachment effectiveness in this case
• the yaw stop & bent claw detachment visually seems to be much better for a full platform climbing (see videos #1,#2)

### Summary #2 - 8/17/04 Testing

On this day, various foot configurations were examined for 4 different gaits.

Notes
• each group contains 3 numbered tests with identical settings (ie. 1,2,3 for yaw stop & 1 bent claw, 13,14,15 for no changes)
• shows the effect of various foot design changes on 4 standard gaits
• the effectiveness of the yaw stop, bent claws are seen
• bending 1 claw has greater effect than just adding yaw stop (MAIN POINTS HERE)
• 1 or 3 bent claws and yaw stop work fairly well together
• more lucid notes may follow as they appear

-- AMcClung? - 15 Oct 2004

Copyright &© by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback