Skip to content

tesseract_robotics

car_seat_example, tesseract_robotics 0.35.0.9.dev534

Python bindings for Tesseract, the ROS-Industrial motion planning framework. Load a robot, check collisions, solve kinematics and plan collision-free trajectories from Python, without ROS.

tesseract_robotics binds the Tesseract C++ libraries module by module: the environment and scene graph, kinematics (KDL, OPW, UR), collision checking (Bullet, FCL), the OMPL, TrajOpt, TrajOpt-Ifopt and Descartes planners, time parameterization and the task composer that chains them into pipelines. On top sits a short Python layer: a planning API, a browser viewer that plays trajectories, and emitters that write controller programs (ABB RAPID, KUKA KRL, Fanuc TP, Yaskawa INFORM, URScript). One wheel per platform covers Python 3.12 and later.

Plan and watch a trajectory

Bash
pip install --pre tesseract-robotics-nanobind
Python
from tesseract_robotics import get_tesseract_resource_path
from tesseract_robotics.common import GeneralResourceLocator
from tesseract_robotics.planning import (
    CartesianTarget, JointTarget, MotionProgram, Pose, Robot, plan_freespace,
)
from tesseract_robotics.viewer import TesseractViewer

locator = GeneralResourceLocator(paths=[get_tesseract_resource_path()])
robot = Robot.from_tesseract_support("abb_irb2400", locator)

# All-zero joints (rad) to a tool-down pose of tool0 in base_link: metres, [qx, qy, qz, qw]
program = (
    MotionProgram("manipulator", tcp_frame="tool0", working_frame="base_link")
    .move_to(JointTarget([0.0] * 6))
    .move_to(CartesianTarget(Pose.from_xyz_quat(0.8, -0.2, 0.8, 0.0, -1.0, 0.0, 0.0)))
)
result = plan_freespace(robot, program)
assert result.successful, result.message

with TesseractViewer() as viewer:
    viewer.update_environment(robot.env)
    viewer.update_trajectory(result.raw_results)
    input(f"{len(result.trajectory)} states, {result.trajectory[-1].time:.2f} s: open {viewer.url}, Enter quits")

plan_freespace runs the TrajOptPipeline: TrajOpt, a discrete collision check of every state, and time parameterization. result.trajectory holds joint positions (rad), velocities, accelerations and time stamps (s); the viewer plays it in the browser.

Next steps

  • Install


    Wheels for Linux, macOS and Windows, or a source build with pixi.

    Installation

  • Quickstart


    Kinematics, a collision check and three planners on one robot, in ten minutes.

    Quickstart

  • Concepts


    The environment, programs, pipelines and profiles, and the frames a target names.

    Concepts

What is in the package

Task Guide Modules
Load and edit a robot cell Environment and scene graph environment, scene_graph, urdf, srdf
Forward and inverse kinematics, Jacobians Kinematics kinematics, state_solver
Distances and contacts, one state or a swept motion Collision checking collision, geometry
Plan freespace, Cartesian and raster motions Motion planning planning, motion_planners_*
Chain planners into pipelines Task composer task_composer, task_composer_planning
Replan in a loop, add constraints in Python Low-level SQP trajopt_ifopt, trajopt_sqp
Watch and record trajectories Viewer viewer
Write controller programs Program emitters emitters
Save environments and programs Serialization serialization

Every recorded example plays in the browser on the examples pages.

Versions

A release 0.A.B.C wraps the Tesseract A.B series; C counts the binding releases. The 0.36 series builds tesseract, trajopt and tesseract_planning from upstream master plus the upstream pull requests it carries; tesseract_robotics.build_identity.UPSTREAM_SOURCES names the exact commits. Changes are in the changelog and the breaking API changes.