Torobo humanoid robot accurately hammers nail, shows potential for industrial use

A humanoid robot developed by Tokyo Robotics is seen accurately hammering a nail in a piece of wood in a new video. The synchronised and accurate effort by Torobo shows its potential for industrial use, according to the manufacturer.

With a height of 1660 mm, a reach of 680 mm, the size of the robot is close to that of humans. According to the company, the payload of one arm is 8 kg at the worst-case holding posture and 20 kg at the instantaneous maximum holding force (when the elbow is bent at a right angle), which are sufficient to conduct robotics R&Ds to replace human work.

The company stated that the action performed by robot is easier than it looks.

“The instantaneous rebound force from the hammer is absorbed through a combination of the elasticity of the rubber material securing the hammer, the deflection in torque sensors and harmonic gears, backdrivability, and impedance control,” said Tokyo Robotics.

“This allows the nail to be driven with a certain amount of force.”

Force-controlling task executions

Equipped with torque sensors at all joints of the arms and waist, the joint torque control enables safe contact stops and force-controlling task executions. Assembly tasks, cooking, and physical interaction with humans can be performed more safely and skilfully, according to the company .

Tokyo Robotics further maintained that the joint configuration of 7-axis dual arms, 3-axis waist (pitch, pitch, yaw), 3-axis neck (yaw, pitch, roll), and 4-axis undercarriage (omni-directional mobile base) enables the robot to do tasks with a range of motion similar to human beings in living and working spaces of people.

Torobo can be used for research on automation of tasks

Developed to accelerate research into industrial applications of full-body humanoid robots, Torobo can be used for research on automation of tasks that involve active contact with people, the environment, and objects, research on applications for next-generation force-controllable dual-armed robots , and research on applying machine learning to robots.

Torobo can utilize impedance control in a Cartesian coordinate system, fall prevention by monitoring ZMP, a state machine for connecting multiple movements, and safety stop based on interference detection.

Torobo’s software is based on ROS. Therefore, in addition to the easy use of the above functions, state visualization with RViz, trajectory planning using a ROS standard software MoveIt!, and logging and saving of the robot’s sensor information (camera images, joint angles, joint torques, etc.) are available.

The robot uses Noitom’s motion capture system, Perception Neuron PRO. This system is used for motion teaching and machine learning as well as remote control applications.

The system provides singularity avoidance, self-interference protection, and a torque-limiting function when the robot is physically constrained (e.g., when pushing a desk from above). It also has functions to enable/disable control of the left and right arms, grippers, neck, waist, and mobile base independently, according to the company .