Ola Shorinwa,Matthew Devlin,Elliot W. Hawkes,Mac Schwager
Abstract:We present DisCo, a distributed algorithm for contact-rich, multi-robot tasks. DisCo is a distributed contact-implicit trajectory optimization algorithm, which allows a group of robots to optimize a time sequence of forces to objects and to their environment to accomplish tasks such as collaborative manipulation, robot team sports, and modular robot locomotion. We build our algorithm on a variant of the Alternating Direction Method of Multipliers (ADMM), where each robot computes its own contact forces and contact-switching events from a smaller single-robot, contact-implicit trajectory optimization problem, while cooperating with other robots through dual variables, enforcing constraints between robots. Each robot iterates between solving its local problem, and communicating over a wireless mesh network to enforce these consistency constraints with its neighbors, ultimately converging to a coordinated plan for the group. The local problems solved by each robot are significantly less challenging than a centralized problem with all robots' contact forces and switching events, improving the computational efficiency, while also preserving the privacy of some aspects of each robot's operation. We demonstrate the effectiveness of our algorithm in simulations of collaborative manipulation, multi-robot team sports scenarios, and in modular robot locomotion, where DisCo achieves $3$x higher success rates with a 2.5x to 5x faster computation time. Further, we provide results of hardware experiments on a modular truss robot, with three collaborating truss nodes planning individually while working together to produce a punctuated rolling-gate motion of the composite structure. Videos are available on the project page: <a class="link-external link-https" href="https://disco-opt.github.io" rel="external noopener nofollow">this https URL</a>.
What problem does this paper attempt to address?
This paper attempts to address the issue of rich contact interactions in multi-robot collaboration. Specifically, it aims to develop a distributed algorithm that enables multiple robots to jointly complete tasks by optimizing forces in a time sequence, such as collaborative manipulation, robot team sports (e.g., robot soccer and robot hockey), and modular robot movements. These tasks typically involve complex contact interactions between robots and objects or the environment, and existing distributed solutions often ignore or simplify these contact interactions, resulting in simplified models that cannot fully utilize the rich contact interaction space, thereby limiting the dexterous manipulation capabilities of the robots.
### Main Contributions of the Paper:
1. **Proposing the DisCo Algorithm**: This is a distributed algorithm for solving multi-robot trajectory optimization problems, particularly suitable for tasks involving rich contact interactions. DisCo leverages the inherent separable structure of the optimization problem, providing efficient local iterations for each robot through the ADMM framework.
2. **Improving Computational Efficiency and Success Rate**: Compared to centralized solvers, DisCo improves average solving time by 150% to 400%, with a success rate exceeding 200%.
3. **Extensive Application Validation**: The paper demonstrates the effectiveness of DisCo in collaborative manipulation, robot team sports, and modular robot movements through simulations and hardware experiments. Specifically, it includes:
- Simulation Experiments: DisCo performs excellently in various multi-robot manipulation tasks, particularly in pushing tasks, achieving a success rate over 3 times higher and reducing computation time by more than 5 times.
- Hardware Experiments: DisCo plans motion gaits on a soft modular truss robot platform, showcasing its feasibility in practical applications.
### Problems Addressed:
- **Complex Contact Interactions**: Existing methods often ignore or simplify contact interactions, failing to fully utilize the rich contact interaction space. DisCo addresses this issue through distributed optimization, enabling robots to perform dexterous operations in complex environments.
- **Computational Efficiency**: Centralized solvers are inefficient in handling large-scale optimization problems. DisCo significantly improves computational efficiency by decomposing the problem into multiple small-scale local problems.
- **Heterogeneous Robot Collaboration**: DisCo supports collaboration among heterogeneous robots with different modalities, such as push-pull robots, aerial robots, mobile manipulators, and fixed-base manipulators working together.
### Conclusion:
The DisCo algorithm effectively addresses the trajectory optimization problem involving rich contact interactions in multi-robot systems through a distributed optimization approach, improving computational efficiency and task success rate, and has been validated in various application scenarios.