Learning Decentralized Multi-Biped Control for Payload Transport

Collaborative Robotics and Intelligent Systems (CoRIS) Institute
Oregon State University

Abstract

Payload transport over flat terrain via multi-wheel robot carriers is well-understood, highly effective, and configurable. In this paper, our goal is to provide similar effectiveness and configurability for transport over rough terrain that is more suitable for legs rather than wheels. For this purpose, we consider multi-biped robot carriers, where wheels are replaced by multiple bipedal robots attached to the carrier. Our main contribution is to design a decentralized controller for such systems that can be effectively applied to varying numbers and configurations of rigidly attached bipedal robots without retraining. We present a reinforcement learning approach for training the controller in simulation that supports transfer to the real world. Our experiments in simulation provide quantitative metrics showing the effectiveness of the approach over a wide variety of simulated transport scenarios. In addition, we demonstrate the controller in the real-world for systems composed of two and three Cassie robots. To our knowledge, this is the first example of a scalable multi-biped payload transport system.

Our goal

Our goal is to transport a payload using an arbitrary carrier, which can be commanded about any control point on the carrier.

Interpolate start reference image.

Our approach

We design a decentralized Multi-Biped Controller (decMBC) where:

  • Each robot independently maneuvers the carrier.
  • There is no inter-communication between robots.
  • Each robot is aware only of its relative position with respect to the control point.
  • The same policy can be deployed across any number of robots.
Interpolate start reference image.

Training in simulation

We train with limited carrier configurations (upto 3 robots) simultaneously by varying carrier control point, robots' relative position, carrier / payload masses.

Testing in simulation

We test our policy across variety of carrier configurations, with arbitrary number of robots, and various carrier and payload weights.

Testing on real hardware

Testing maneuverability

We test our policy on real robots where same policy is deployed to each of the three robots. We demonstrate maneuverability of the carrier by commanding the carrier to follow different user commands.

Testing robustness to disturbances

We also test the robustness of our policy against disturbances by applying external forces to the carrier. Here, we deploy the same policy to each of the two robots.

More hardware tests!

We test our policy for two and three robots with various carrier, attachment points for robots, and payload weights.

BibTeX

@misc{pandit2024learningdecentralizedmultibipedcontrol,
        title={Learning Decentralized Multi-Biped Control for Payload Transport},
        author={Bikram Pandit and Ashutosh Gupta and Mohitvishnu S. Gadde and Addison Johnson and Aayam Kumar Shrestha and Helei Duan and Jeremy Dao and Alan Fern},
        year={2024},
        eprint={2406.17279},
        archivePrefix={arXiv},
        primaryClass={cs.RO}
        url={https://arxiv.org/abs/2406.17279},
}