The projects main goal is to increase the robustness of biped walking on uneven terrain and floors with different mechanical properties. We evaluate the performance of the developed methods by walking terrain-blind over a set of unknown obstacles --- i.e. no vision-based information is required for the approaches to work. The focus is on the fast (≥ 0.5 m/s) traversal of such environments.
Contact Force Control
The robustness to uncertain and undetected terrains is achieved by control of the ground reaction forces of the biped robot. Our control scheme includes reactive feedforward components for early contacts of the feet, as well as feedback control of the sensed contact wrenches. The force control approach considers sensor-based geometric contact information of the contact to the ground as well as the dynamics of the center of mass.
More information is available in our publication.
Currently, the contact state and estimated contact geometry (pressure distribution) between the foot and the ground is sensed via four discrete contact switches on each foot. To get a more accurate information on the contact geometry in uneven terrain, a lightweight and flexible tactile sensor has been developed. By integration of the sensor to the robot's feet, we expect increased performance of our existing force-control approaches in uneven terrain.
For more information, please see our publication:
A Flexible and Low-Cost Tactile Sensor for Robotic Applications
The approaches are validated in a complex scenario with uneven, irregularly protruded surfaces. All tests are conducted without any (vision-based) information on the terrain. This means the motion plan is generated for flat terrain and only the force-control may adapt the trajectories based on the sensor data.
Several videos of experiments related to the contact force control of LOLA are available online:
The humanoid robot LOLA walks over uneven terrain without using visual information
A Force Control Scheme for Biped Robots to Walk over Uneven Terrain Including Partial Footholds
Hybrid Position/Force Control for Biped Robot Stabilization with Integrated Center of Mass Dynamics