Planning and Control of a Humanoid Robot for Pushing a Heavy Object during Locomotion

Humanoid robots have the potential to replicate human skills in both locomotion and manipulation. For this reason, they can be employed to support humans in real-life daily tasks. Pushing large and heavy objects during locomotion is certainly one of the most important skills that is expected to be achieved since it is common in both industrial and domestic environments. However, friction, non-linearities, high interaction forces and robot’s intrinsic limits make this task difficult to achieve. Based on existing methods for humanoid walking, this thesis extends the walking planning algorithm based on the Divergent Component of Motion (DCM) by considering the necessary external force that the robot should apply to push and move a heavy object according to a desired trajectory. This new planning algorithm is then combined with inverse dynamics whole-body control to allow the humanoid robot to perform the pushing task while walking. The proposed control framework is tested in simulation on BRUCE, a kid-size humanoid robot. Results show that the proposed framework allows the robot to carry an object whose weight is approximately half the weight of the robot by pushing it with its hands.