projects

projects store robot definitions and assets under reusable paths such as Robot('my-project/my-robot') and Asset('my-project/cup.glb'). sign in and open projects to create or manage one.

import and configure

use build a robot to upload a definition and its meshes, or paste a github url to a urdf, mjcf, or xacro definition. glb and step imports are also available; the import api accepts fusion design data.

edit the part tree, joints, default poses, collision geometry, and floating-base setting. configure named servos, motors, encoders, imus, cameras, drives, and arms; these names become the python subsystem keys. gallery robots can be copied into a project before editing.

an arm selects an ordered servo chain and a tool-tip part or marker. its kinematics come from the same robot geometry used by the simulator and viewer. move tip markers in the 3d editor to set the tool frame, then check it with arm.describe() in a running scene.

assets

the assets tab uploads, replaces, renames, downloads, and deletes project files. glb uploads generate a collision mesh; switch between visual and collision to inspect both. other files, including onnx policies, are stored without a 3d preview.

from botblocks import Asset, OnnxInference

cup = Asset('my-project/cup.glb')
policy = OnnxInference('my-project/policy.onnx', inp_len=61, pred_len=14)

the policy dimensions must match your model. see the python api for assets and inference.

sharing

project settings controls the project's public/private visibility and name. public projects can be read by other users; editing remains owner-only. private robot and asset loads require your signed-in credentials. use botblocks login for local scripts.

renaming a project, robot, or asset changes its path, so update scripts that load it. deleting a project removes its robots and assets too.