Hardware Design and Learning-Based Software Architecture of Musculoskeletal Wheeled Robot Musashi-W for Real-World Applications (2403.11729v1)

Abstract: Various musculoskeletal humanoids have been developed so far. While these humanoids have the advantage of their flexible and redundant bodies that mimic the human body, they are still far from being applied to real-world tasks. One of the reasons for this is the difficulty of bipedal walking in a flexible body. Thus, we developed a musculoskeletal wheeled robot, Musashi-W, by combining a wheeled base and musculoskeletal upper limbs for real-world applications. Also, we constructed its software system by combining static and dynamic body schema learning, reflex control, and visual recognition. We show that the hardware and software of Musashi-W can make the most of the advantages of the musculoskeletal upper limbs, through several tasks of cleaning by human teaching, carrying a heavy object considering muscle addition, and setting a table through dynamic cloth manipulation with variable stiffness.

• The paper demonstrates the integration of bio-inspired musculoskeletal arms with a mechanum-wheeled base for versatile, adaptive task performance.
• It details a hybrid software architecture combining static and dynamic body schema learning with reflex controls for precise movement and safe operation.
• Experimental results, including cleaning, heavy object transport, and cloth manipulation tasks, highlight the robot's practical utility in real-world environments.

Musculoskeletal Wheeled Robot Musashi-W: Integrating Flexible Limbs with Wheeled Mobility for Real-World Applications

Hardware Overview

The Musashi-W robot combines musculoskeletal upper limbs with a wheeled base, incorporating both flexible, bio-inspired arm and hand structures and the stability and mobility provided by mechanum wheels. The dual arms feature joint modules, aluminum frames, and various muscle actuators, including integrated sensor-driver muscle modules and miniature bone-muscle modules. These components allow for an adaptable musculature configuration, supporting redundancy and the addition of muscles for task-specific enhancements. The wheeled base utilizes four mechanum wheels, a linear motion mechanism, and a comprehensive circuit system for robust movement and task execution. The design intricacies balance the complexities of a musculoskeletal system with the practicalities of a wheeled robot, aiming for versatile real-world applications.

Software Architecture

The software system underpinning Musashi-W merges static and dynamic body schema learning, reflex control mechanisms, and traditional controls for movement and task operations. Static body schema learning provides a fundamental understanding of the robot's body mechanics, enabling motion control through muscle actuation. Dynamic body schema learning further refines control by accounting for interactions with tools and objects, adjusting for nuanced tasks execution. Reflex controls, including muscle relaxation and thermal management systems, ensure the robot's musculature operates within safe parameters. Additionally, classical controls manage the wheeled base and mechanum wheels for precise movement.

Experimental Demonstrations

Duster Experiment

Through human demonstration using a VR controller, Musashi-W successfully executed a cleaning task, maneuvering to a desk, grasping a duster, and dusting a shelf. This experiment showcased the robot's capability to integrate complex movements and adapt to real-world environments, benefiting from its software and hardware design.

Heavy Object Carrying

Musashi-W demonstrated its task-specific adaptability by carrying a heavy object, utilizing added muscles and relearned static body schemas to perform the task efficiently. By increasing the muscle arrangement complexity, the robot could reduce the maximum muscle tension required, highlighting the benefits of its musculoskeletal design.

Table Setting with Dynamic Cloth Manipulation

In a table setting task, Musashi-W displayed sophisticated dynamic manipulation, adjusting cloth over a table by altering its joint angles and stiffness. The robot's ability to execute such delicate tasks underlines the efficacy of the dynamic body schema learning and its implementation in real-world applications.

Conclusion and Future Prospects

The development of Musashi-W represents a significant step toward the integration of musculoskeletal flexibility and wheeled mobility in robotics, particularly for complex, real-world tasks. The comprehensive software architecture enables the robot to learn and adapt to various tasks dynamically. Future developments will likely focus on further enhancing the robot's learning capabilities, generalizability to more tasks, and autonomous decision-making, broadening the scope of real-world applications for musculoskeletal wheeled robots.