Aloha Mini 2 Pro

Aloha Mini 2 Pro#

AlohaMiniCfg loads the complete AlohaMini/alohamini2pro.urdf asset without renaming its joints or assembling separate arms.

Aloha Mini 2 Pro with its complete base, vertical column and two bent arms, rendered with soft studio lighting
Aloha Mini 2 Pro · 15 DOFs across five control groups

The image uses the registered AlohaMini preset in a display pose, rendered with RenderCfg(renderer="rt") (OfflineRT). Three soft area lights reveal the dark arm details while retaining the white body’s surface definition.

Image rendering settings

960 × 1200; 1 sample per pixel; DLSS offscreen reconstruction (quality 2); tone mapping off; exposure 1.0. The default ground and environment emission are supplemented by neutral key, fill and rim rectangular lights. Exact camera, lighting and joint settings are recorded in the render manifest.

from embodichain_robots import AlohaMiniCfg

cfg = AlohaMiniCfg.from_dict({"uid": "aloha_mini"})
robot = sim.add_robot(cfg=cfg)

Control part

DOF

Joints / endpoint

left_arm

6

Shoulder through wrist; left_tcp

right_arm

6

Shoulder through wrist; right_tcp

left_eef

1

left_gripper

right_eef

1

right_gripper

torso

1

vertical_move

The arms use SSIKSolverCfg(backend="warp") on CPU or CUDA with the TCP frames authored in the asset. The torso retains its position-only PytorchSolverCfg; the hands use joint-space commands. Explicit solver replacements remain supported; see the original numerical solver settings. The source’s three planar base joints and three wheel joints remain outside these control parts, with position drives that hold their initial positions. Drive gains follow the CobotMagic simulation preset and can be overridden through joint_drive_props.

cfg.build_pk_serial_chain() returns all five parts. Each chain is local to its part: arm poses are relative to left_Base / right_Base, hand poses to the fixed jaws, and torso poses to base_cad_link. Arm chains do not include torso or planar base motion. Hand chains describe moving-jaw poses, while the arm chains describe the grasp TCPs.

Run a headless construction smoke check with:

python -m embodichain_robots.alohamini.aloha_mini.cfg --headless --device cpu

Omit --headless for the interactive simulation window. Select the physics backend with --physics default or --physics newton.

See the API reference.