Dynamic Legged Manipulation of a Ball Through Multi-Contact Optimization

Chenyu Yang,Bike Zhang,Jun Zeng,Ayush Agrawal,Koushil Sreenath

DOI: https://doi.org/10.48550/arXiv.2008.00191

2020-12-21

Abstract:The feet of robots are typically used to design locomotion strategies, such as balancing, walking, and running. However, they also have great potential to perform manipulation tasks. In this paper, we propose a model predictive control (MPC) framework for a quadrupedal robot to dynamically balance on a ball and simultaneously manipulate it to follow various trajectories such as straight lines, sinusoids, circles and in-place turning. We numerically validate our controller on the Mini Cheetah robot using different gaits including trotting, bounding, and pronking on the ball.

Robotics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to make a quadruped robot dynamically balance on a sphere while manipulating the sphere to move along a specified trajectory through multi - contact optimization. Specifically, the paper proposes a model predictive control (MPC) framework, enabling the quadruped robot to dynamically balance on the ball and simultaneously manipulate the sphere to move along various trajectories such as straight lines, sine waves, circular paths or rotate in - place. The Mini Cheetah robot was used in the study and numerically verified on the ball through different gaits (such as trotting, jumping and pouncing).

### Core Problems of the Paper
1. **Dynamic Legged Manipulation**: How to design a feedback controller to enable the quadruped robot to interact with objects.
2. **Manipulation of Unactuated Objects**: The objects to be manipulated are usually unactuated, which increases the under - actuation degree of the quadruped robot.
3. **Reference Trajectory Tracking**: Besides interacting with objects, the robot also needs to make the objects move along the reference trajectory or reach the desired pose.
4. **Manipulation under Contact Constraints**: The manipulation of objects is achieved through the movement of the robot, and the movement of the robot is restricted by the unilateral and frictional contact constraints between the robot and the objects.
5. **Comprehensive Challenges**: The problem combines the challenges of legged locomotion (including hybrid and nonlinear dynamics and high under - actuation degrees) and the challenges of non - grasping manipulation.

### Solutions
1. **System Modeling**: Decouple the entire system through contact points and systematically describe the task of the quadruped robot dynamically manipulating the sphere.
2. **Interaction Model**: Based on the concept of equivalent generalized mass, develop a simplified interaction model between the quadruped robot and the sphere.
3. **Reaction - Force - Oriented MPC (R - MPC)**: Design an MPC strategy considering contact forces to achieve the dynamic manipulation of the sphere along a given trajectory.
4. **Foot - Placement Controller**: Propose a foot - placement controller based on constrained quadratic programming to adapt to the spherical surface.

### Experimental Verification
The paper verifies the effectiveness of the proposed control strategy through simulation, using the MIT Cheetah software and an elastic contact model to calculate the interaction forces between the robot and the sphere. The experimental results show that the proposed controller outperforms the baseline controller in terms of speed tracking and sphere torque command tracking, especially under different gaits and reference trajectories.

### Main Contributions
1. Systematically describe the task of the quadruped robot dynamically manipulating the sphere.
2. Develop a simplified interaction model between the quadruped robot and the sphere.
3. Design a reaction - force - oriented MPC strategy considering contact forces.
4. Propose a foot - placement controller adapted to the spherical surface.

### Conclusion
The paper proposes a new control design that enables the quadruped robot to effectively manipulate the sphere to move along a specified trajectory while dynamically balancing on the sphere. This method not only expands the manipulation range of the robot but also improves its flexibility and agility on extreme terrains.