Abstract:
The proposed thesis work focuses on the design and control of a new unique agile and dexterous mobile manipulator, the Carnegie Mellon University (CMU) ballbot. The CMU ballbot is a human-sized dynamically stable mobile robot that balances on a single ball. We present the development and integration of a new pair of seven-degree-of-freedom (7-DOF) humanoid arms to the CMU ballbot. To the best of our knowledge, this robot configuration is the first and only of its kind. The underactuated ballbot class of robots presents unique challenges in planning, navigation, and control; however, it also has significant advantages over conventional mobile robots. The new multi-DOF arms add to the existing complexities of ballbots. Careful coordination between the upper and lower body to maintain balance while performing manipulation tasks is required.

This thesis seeks to demonstrate that the highly dynamic ballbot with 7-DOF arms is controllable over a wide envelope of possible configurations. Within this controllability envelope, we strive to show that purposeful actions can be successfully planned and carried out. We explore the use of centroidal dynamics, which has recently become a popular approach for designing balancing controllers for humanoid robots. As proof of feasibility, we first develop a center of mass compensation controller to enable the ballbot with a pair of 2-DOF arms to lift and transport heavy payloads of up to ~15kg. We later extend this controller to the ballbot with 7-DOF arms.

To enable performing dynamic whole-body motions, we present a momentum-based planning and control framework. We describe a framework where we first solve a trajectory optimization problem offline and later use the same NLP with a shorter time horizon in a model predictive control (MPC) context to execute the motion. We define balancing for a ballbot in terms of the centroidal momentum instead of other approaches like ZMP or angular velocity that are more commonly used.

The remaining work for this thesis will revolve around developing the skills necessary to realize the task of pushing a manual wheelchair with the physical ballbot hardware. We plan to extend our planning and control framework to enable body and end-effector forces planning and control. This will allow the robot to generate force inputs to maneuver the wheelchair. Further, we will integrate vision and grasp planning into the pipeline. A significant part of the proposed work consists of experimental validation on the ballbot hardware. We will challenge the ballbot to achieve different loco-manipulation tasks.

More Information

Thesis Committee Members:
Ralph Hollis, Chair
Nancy Pollard
Oliver Kroemer
Patrick Wensing, University of Notre Dame