What problem does it solve?
Controlling a robot arm to reach, grasp, transport, and place objects in Isaac Sim requires choosing between multiple IK stacks, configuring grasp frames, and tuning solver parameters, and mistakes like hardcoded grasp offsets or wrong Jacobian slicing cause silent failures or PhysX explosions.
Core Features & Use Cases
- Differential IK on Articulation: End-effector control via
get_jacobian_matrices with damped-least-squares, pseudoinverse, transpose, or SVD solvers, including virtual-base DOF slicing rules.
- Schema-native IK and named poses:
isaacsim.robot.poser.RobotPoser for solving IK, storing reusable named poses like "pick_position", and exporting/importing pose libraries.
- Obstacle-aware motion and grasping: cuMotion
RmpFlowController and PINK PinkIKController for reactive trajectories, plus FixedJoint and SurfaceGripper grasp patterns with a three-checkpoint validation loop.
- Use Case: Build a Franka pick-and-place pipeline that discovers the grasp frame, solves IK for approach and placement, attaches the object with a computed FixedJoint offset, and validates grasp, lift, and place success.
Quick Start
Use the manipulation-ik skill to set up differential IK control for a Franka arm to pick and place an object in Isaac Sim.