RoboBoom
Hannes Hoeppner· April 10, 2012· 39:29

Talk on "Soft Robotics Design and Control" by Alin Albu-Schäffer Part2

Alin Albu-Schäffer, a robotics researcher at DLR, explains compliant robot control, safety, and variable-stiffness actuation, with examples from DLR robot systems and collaborative projects.

In this talk

  1. 0:01

    Decoupling Cartesian impedance

    Albu-Schäffer describes using simultaneous diagonalization of stiffness and inertia matrices to separate a robot’s behavior by direction. Demonstrations show low or zero stiffness in selected directions, stable quaternion-based orientation control, and a singularity-avoidance potential that prevents entry into singular configurations.

  2. 9:05

    Passivity for flexible-joint robots

    For flexible joints, motor and link positions and velocities must be modeled separately, making torque part of the system state. Albu-Schäffer describes feedback that reduces effective motor inertia and friction, and a passivity-based framework intended to keep the robot stable despite modeling errors or unknown contact dynamics. He cites a comparison in which the lightweight robot outperformed a 250-kilogram KUKA carrying the same 14-kilogram load in three of five test categories.

  3. 17:08

    Safety, assembly, and whole-body interaction

    The control approach extends to grasping, object-based impedance, and a 33-degree-of-freedom mobile system weighing 150 kilograms. For assembly, vision provides an approximate target and compliance limits contact forces; stability analysis can help choose stiffness and motions that work across a larger range of starting positions. Link-side torque sensing also supports collision detection within one millisecond and helps distinguish contact forces from gearbox friction.

  4. 26:01

    Variable compliance and energy storage

    Albu-Schäffer presents mechanical compliance as useful for human interaction, impact tolerance, and storing energy for dynamic motion. A human-sized hand has 19 degrees of freedom and 38 actuators, with stiffness adjusted through tendon and spring mechanisms; the team is extending passivity-based control to these nonlinear, coupled systems. In a throwing example, storing spring energy lets a robot throw an object several meters, compared with about 0.8 meters without that stored energy.

Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.