Skip to content
WE BUILDHUMANOIDS
photograph pending · front
Image: not yet available1 / 1
Status: SHIPPING

PAL Robotics KANGAROO

PAL RoboticsHumanoid robot

KANGAROO is PAL Robotics' current legged-robot platform for dynamic locomotion, reinforcement-learning and embodied-AI research, sold in configurable variants: 1.60 m tall, 50-65 kg, with 14-40 total DOF and single-arm options of 4, 5 or 7 DOF (each with its own rated payload). Its software stack runs Ubuntu LTS with ROS 2 LTS and PAL OS, using ros2_control for actuator control (current, speed, position, impedance and optional force/torque modes) and provides URDF and MJCF robot models. PAL's own GitHub repositories (kangaroo_robot, kangaroo_simulation, kangaroo_moveit_config) default to the humble-devel branch and are officially released into the ROS 2 Humble distribution; simulation is primarily via MuJoCo (the kangaroo_simulation README names mujoco_ros2_control as 'the primary supported simulator'), and the product page additionally lists mjlab as a supported RL tool. PAL publishes an open-source RL policy library covering walking, push recovery, running, jumping, stair climbing, dance and box picking, plus open-source reinforcement learning, motion retargeting and imitation learning pipelines. A 15 Ah / 63 V (976 Wh) battery and EtherCAT 2 kHz real-time control loop round out the hardware, with mobility rated 'up to 2 m/s' on the datasheet.

EDITORIALOverview reviewed against the cited sources by WBH, .

HEIGHT
1.60 m
WEIGHT
50-65 kg
DOF
14-40 DOF
PAYLOAD
7 kg
Suggest an edit

Last verified14 sources

DATA · 24 FIELDS

Specifications

  • Official source
  • Community measured
  • Not published
PHYSICAL
Height1.60 m1,60 m as published (comma decimal separator); 1600 mm is exact arithmetic conversion.Official source [2]
Weight50-65 kgDepends on platform configuration.Official source [2]
Degrees of freedom14-40 DOFDepends on platform configuration.Official source [2]
Leg DOF6 DOF (per leg)Official source [2]
Arm DOF4, 5 or 7 (configurable)Official source [2]
Torso DOF2 DOF (yaw and roll)Official source [2]
LocomotionBipedalOfficial source [3]
Walking speedup to 2 m/sPublished under 'Mobility'; not stated whether this is walking or running speed, or under which configuration.Official source [3]
Ingress protectionNot publishedNot published
JOINTS & ACTUATION
Max joint torqueNot publishedPAL publishes actuator/force-sensing features (custom linear actuators with integrated force sensors), not a maximum torque figure in N·m.Not published
Control loop frequency2 kHz (EtherCAT)Official source [2]
POWER
Battery976 Wh (15 Ah, 63 V)Official source [2]
Runtime~3 h autonomyDatasheet states '3 hrs autonomy' without specifying walking vs. stand-by mode (unlike TALOS, which publishes both separately).Official source [3]
Charging timeNot publishedNot published
COMPUTE & SENSING
Compute2x Intel i7 (32GB/1TB), optional NVIDIA Jetson GPU; Ubuntu LTS + ROS 2 LTSOfficial source [2]
Perception sensors4x RGB-D (D435i), IMU/AHRS 1 kHz, F/T sensors (wrist/feet)Official source [2][3]
Connectivity2x WiFi6, USB-3/USB-C, 2x Gigabit Ethernet RJ45Official source [2]
Safety / emergency-stop systemsOnboard + wireless E-stop, SIL 2 / PL d ratedOfficial source [2]
HANDS & PAYLOAD
End-effectorParallel gripper or 5-finger hand (RH8D / RH56F1-E2), quick-change flangeOfficial source [2]
Arm payload (4-DOF arm, single-arm extended)7 kg (4-DoF arm, single/extended) · 30 kg (4-DoF arm, bi-manual) · 5 kg (5-DoF arm, single/extended) · 28 kg (5-DoF arm, bi-manual) · 3 kg (7-DoF arm, single/extended) · 25 kg (7-DoF arm, bi-manual)Arm fully extended, single arm; 4 DOF arm configuration.Official source [2]
Hand DOF (parallel gripper)1 DOF (2 fingers) (Parallel Gripper) · 7 DOF (5 fingers) (RH8D) · 6 DOF (5 fingers) (RH56F1 / RH56E2)Seed Robotics parallel gripper option.Official source [2]
AVAILABILITY & PRICE
Indicative priceNot publishedNot published
AvailabilityRequest a quoteOfficial source [2]
WarrantyNot publishedDatasheet's Support Services list includes 'Warranty extension' as an add-on, implying a base warranty exists, but no duration or terms are published.Not published

DATA · SOFTWARE & SIMULATION

Development

INTERFACESUPPORTPACKAGE / VERSIONSRC
SDKNot publishedNo separate proprietary SDK found; KANGAROO is programmed through ROS 2 / ros2_control (see the ROS 2 entry), with a documented 'ROS 2 API C++/Python'.[2]
ROS 2Officialkangaroo_robot (2.15.1/2.16.0), kangaroo_simulation (2.7.0) and kangaroo_moveit_config (2.2.2) are PAL's own GitHub repositories with humble-devel as the default (and, for kangaroo_robot, only non-fork) branch, ROS 2 Humble, bloom-released into the official ROS 2 Humble rosdistro index. No entries exist in the Jazzy (ROS 2) or Noetic (ROS 1) rosdistro indices; unlike TALOS, KANGAROO's repositories never had a ROS 1 branch, so ROS 2 Humble is the only distro it has ever officially supported.[2][5][10]
Isaac Sim / LabNot publishedNo official or community Isaac Sim/Lab package or environment found for KANGAROO in the sources checked.
MuJoCoOfficialkangaroo_simulation repository (default branch humble-devel) states MuJoCo (via mujoco_ros2_control, in the kangaroo_mujoco package) is 'the primary supported simulator', officially released as kangaroo_mujoco 2.7.0 in the ROS 2 Humble rosdistro. The product page and datasheet additionally list 'MuJoCo and mjlab' under Simulation and RL Tools; PAL also maintains a public mujoco_vendor ROS 2 vendor package and a pal_mjlab repository.[2][6][10]
GazeboNot publishedNo official Gazebo package, dependency or mention found for KANGAROO; the kangaroo_simulation README and product page name only MuJoCo/mjlab as supported simulators. Not marked 'none' because no PAL source explicitly states Gazebo is unsupported.
URDF / MJCFOfficialkangaroo_robot repository includes a kangaroo_description package (URDF); product page states the 'Robot Model' is 'URDF and MJCF Model'.[2][4]
MoveIt!Officialkangaroo_moveit_config package (v2.2.2, humble-devel) provides a MoveIt! configuration for KANGAROO, per the ROS 2 Humble rosdistro index.[7][10]
RL policy libraryOfficialPAL publishes a named RL Policies Library for KANGAROO: Walking, Push recovery, Running, Jumping, Stair climbing, Dance, Box picking; plus open-source Reinforcement Learning, Motion Retargeting and Imitation Learning pipelines and an 'AI Controller' for online policy switching.[2]
LanguagesOfficialkangaroo_robot repository: Python 96.3%, CMake 2.1%, C++ 1.6% (GitHub language statistics).[4]