Action planning

Use task_type act with a target goal to return action-ready spatial guidance.

What it answers

What sequence of moves gets the scene to the goal state?

task_type: act

What to send

input.goalobject

The target state to reach.

input.cad_sceneobject

Current scene. point_cloud_scene also works.

input.object_idstring

Optional. The object being acted on.

Every field on the request schema is still accepted; these are the ones this task depends on.

What comes back

spatial_action.steps

Ordered actions grounded in scene geometry.

spatial_action.constraints

Physical limits each step respects.

spatial_action.blocked

Steps that cannot be satisfied, and why.

Example

Terminal

$ curl https://fourechelon.com/api/v1/responses \
    -H "Authorization: Bearer $FOUR_ECHELON_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
$     "model": "rowe-1.0",
$     "input": {
$       "task_type": "act",
$       "goal": { "state": "assembled" },
$       "cad_scene": { "schema": "four-cad.cad_scene.v1", "solids": [] }
$     }
$   }'

Worth knowing

This is a planning service, not a controller. It runs off-device over the network with no real-time or safety guarantee — treat the plan as a proposal your system validates.