Back to PAL Robotics
KANGAROO

KANGAROO

KANGAROO

Last updated: September 27, 2026
Locomotion Type
Biped
Stage
Mass Production
Height
160 cm
Degrees of Freedom
28 DOF
Weight
50 kg
Max Speed
4.5 km/h
Battery Life
6 h
Release Year
2021
Computing
Control PC and multimedia PC, each Intel i7 with 32 GB RAM and 1 TB SSD; optional NVIDIA Jetson GPU; 2x Wi-Fi 6
Sensors
4x RGB-D cameras; IMU and AHRS at 1 kHz (0.25 deg static, 0.5 deg dynamic); optional 6-axis force/torque sensors at wrist and feet; force sensors integrated in leg actuators

KANGAROO is PAL Robotics' dynamic bipedal humanoid, launched in 2021 and put on sale in 2025 in two versions, KANGAROO and KANGAROO PRO with two arms. It is built as a development-ready platform for research on dynamic locomotion, reinforcement learning and embodied AI, and PAL describes its purpose as bridging the gap between the lab and the real world. The robot stands 1.60 m, weighs between 50 and 65 kg depending on configuration, and offers 14 to 40 degrees of freedom: a 6-DoF leg, a 2-DoF torso for yaw and roll, arms configurable at 4, 5 or 7 DoF, and an ISO 9409-1 arm flange with quick tool changer. Its legs are designed with low moving inertia for highly dynamic motion, and balance is handled by a whole-body inverse dynamics controller; the real-time stack runs on EtherCAT with a 2 kHz control loop, and an IMU and AHRS update at 1 kHz. Perception includes four RGB-D cameras, with optional six-axis force/torque sensors at the wrist and feet and force sensors integrated into the leg actuators. Safety is addressed with onboard and wireless emergency stops plus SIL 2 and PL d ratings. Compute is dual Intel i7 PCs with 32 GB RAM and 1 TB SSD each and an optional NVIDIA Jetson GPU, and the software stack is Ubuntu LTS with ROS 2 LTS and PAL OS on ros2_control, with URDF and MJCF models, MuJoCo and mjlab simulation, an AI controller framework for switching control policies online, and an open-source reinforcement-learning policy library covering walking, push recovery, running, jumping, stair climbing, dance and box picking.

Application Scenarios

Research on dynamic legged locomotion, reinforcement learning and embodied AI; development platform bridging lab research and real-world deployment

Subscribe to Weekly Report

Get weekly updates on humanoid robot funding, new companies, and product launches.