SlashML (/ml) Embodied Robotics 6-DoF Spatial Memory

Query and update persistent 6-DoF SE(3) spatial world state, kinematic object trajectories, and task affordances for embodied robots and ROS 2 nodes via the SlashML Embodied Robotics API and MCP tools.

/ ml (SlashML) Embodied Robotics Spatial Memory Skill

Use this skill when building or controlling embodied robotics agents, VLA policies, ROS 2 nodes (tf2, Nav2, MoveIt 2), or spatial world-state trackers backed by slashml-robotics (@slashml/robotics).

Core Spatial Operations

1. Record 6-DoF Pose & Affordance Observations (record_sensor_observation / POST /v1/robotics/observations): Record observed object poses (pose: [x, y, z, qx, qy, qz, qw]), workstation zones (room_zone), and task affordance tags (tags).

   python

   from slashml_robotics import Pose6D, RoboticMemory
   
   memory = RoboticMemory(api_key="sml_live_...")
   memory.add_observation(
       object_name="torque_wrench",
       room_zone="machining_bay",
       pose=Pose6D(x=1.25, y=-0.40, z=0.82),
       tags=["tool", "fastening"],
       confidence=0.98,
   )
   

2. Query Latest Object Belief State (locate_physical_object / GET /v1/robotics/where_is/{object_name}): Retrieve the current superseded 6-DoF pose, room zone, and confidence for an object without scanning stale historical frames.

   python

   loc = memory.where_is("torque_wrench")
   

3. Retrieve Kinematic Trajectory (query_object_trajectory / GET /v1/robotics/trajectory/{object_name}): Inspect chronological pose observations to trace how an object moved across workspaces over time.

   python

   traj = memory.get_trajectory("torque_wrench")
   

4. Lookup Objects by Affordance (find_objects_by_affordance / GET /v1/robotics/affordance/{action}): Find all tracked objects in the workspace that satisfy a required manipulation or navigation affordance (e.g., fastening, measuring).

   python

   matches = memory.find_by_affordance("fastening", room_zone="machining_bay")