Sequential manipulation planning for over-actuated unmanned aerial manipulators

Yao Su, Jiarui Li, Ziyuan Jiao, Meng Wang, Chi Chu, Hang Li, Yixin Zhu, Hangxin Liu

TL;DR

This paper tackles the complex challenge of enabling Unmanned Aerial Manipulators (UAMs) to perform sequential, multi-step manipulation tasks. They introduce a novel over-actuated UAM platform and a Virtual Kinematic Chain (VKC) based planning framework that unifies the motion of the flying base and the manipulator, enabling coordinated, complex aerial manipulation.

What Problem Does This Paper Address?

Current aerial manipulation research primarily focuses on single-step tasks like pick-and-place. However, many real-world applications require UAMs to perform a series of coordinated actions, such as opening a door, retrieving an object, and then placing it elsewhere. This "sequential manipulation" is significantly more challenging because it demands:

  1. Coordinated Motion: The UAM's flying base and its manipulator must move in a synchronized manner.
  2. Dynamic Adaptation: The UAM needs to adapt its motion patterns for different stages of a task, especially when interacting with objects of varying kinematic structures.
  3. Unified Planning: A planning approach is needed that considers the kinematics and dynamics of the entire system—flying base, manipulator, and the object being manipulated—simultaneously.

Existing UAM platforms often lack the necessary degrees of freedom or the sophisticated planning frameworks to handle these multi-step operations effectively.

Methods

The authors propose a comprehensive solution involving a new UAM design and a specialized planning framework:

  • Over-Actuated UAM Platform: They designed a UAM that integrates a 4-degree-of-freedom (DoF) manipulator with an over-actuated UAV. This UAV is composed of four modular quadcopters, each with a passive gimbal mechanism, allowing for enhanced maneuverability and attitude control. This design provides the necessary agility and redundancy for complex tasks.
  • Virtual Kinematic Chain (VKC) Framework: Inspired by ground-based mobile manipulation, they extend the VKC concept to aerial platforms. A VKC consolidates the kinematics of the UAM's flying base, its manipulator, and the manipulated object into a single kinematic chain. This allows the entire system to be planned as a whole, leading to more coordinated motions.
  • Virtual Base: A virtual base is introduced to model the free motion capabilities of the over-actuated UAV.
  • Virtual Joints: Virtual joints connect the manipulator's end-effector to the object's model, and also link components of the manipulator itself.
  • Trajectory Optimization: The sequential manipulation problem is formulated as finding an optimal trajectory for the VKC. This is achieved through trajectory optimization, minimizing factors like traveled distance and ensuring trajectory smoothness, while satisfying task-specific constraints (e.g., reaching a goal pose) and safety constraints (e.g., collision avoidance).
  • Hierarchical Control Architecture: A two-level control system is employed:
  • High-Level Controller: Runs on a remote PC, responsible for trajectory planning and generating desired commands.
  • Low-Level Controller: Runs onboard the UAM at a higher frequency, executing the commands, managing motor torques, and ensuring precise tracking of desired positions, attitudes, and manipulator joint angles. This includes a downwash-aware control allocation method for efficient thrust management.

Datasets

Not discussed in the paper. The paper focuses on simulation and experimental validation of their proposed framework and platform, rather than training on large datasets.

Key Insights and Ideas

  • VKC for Aerial Manipulation is Key: The paper demonstrates that extending the VKC concept, successful in ground robotics, to aerial platforms is crucial for achieving coordinated sequential manipulation. It effectively unifies the planning of disparate components (base and arm).
  • Over-Actuation is Essential: The proposed UAM's over-actuated nature (more actuators than strictly necessary for basic flight) is highlighted as critical for achieving the agility, redundancy, and precise control needed for complex manipulation tasks.
  • Hierarchical Control is Practical: The combination of a high-level planner and a fast, onboard low-level controller is shown to be effective in tracking complex, dynamically changing trajectories generated by the VKC planner.
  • Sequential Tasks are Now Feasible: The paper moves beyond single-step manipulation, showing that UAMs can be programmed to perform multi-step sequences, opening doors to more sophisticated applications.

Key Results

The paper presents results from both simulation and physical experiments, demonstrating the effectiveness of their approach:

  • Simulation: They successfully simulated two sequential tasks:
  • Task 1 (Light Bulb Installation): A 4-step task involving approaching, picking up, rotating, translating, and inserting a light bulb.
  • Task 2 (Object Relocation): A 6-step task involving approaching a cabinet, opening the door, picking up an object, placing it inside, and closing the door.
  • Experiment: They successfully replicated the object relocation task (similar to Task 2) in a physical experiment. The UAM accurately tracked the reference trajectory generated by the VKC planner, demonstrating the practical viability of their system. Keyframes from both simulation and experiment are provided, visually showcasing the successful execution of these complex sequences.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that the current UAM platform relies on a remote PC for high-level planning, introducing potential communication delays.
  • While the system demonstrates impressive capabilities, the complexity of the tasks is still limited to a few sequential steps.
  • Future Work:
    • Onboard Computing: Integrating more powerful onboard computing to reduce reliance on remote PCs.
  • Wireless Tactile Sensing: Incorporating wireless tactile sensors into the manipulator to enhance manipulation capabilities and feedback.
  • Broader Application Range: Expanding the framework to handle even more complex and longer sequences of aerial manipulation tasks.
  • Reconfigurability: Further exploring the potential of modular and reconfigurable UAM designs for diverse manipulation scenarios.

Summary updated

Abstract

We investigate the sequential manipulation planning problem for unmanned aerial manipulators (UAMs). Unlike prior work that primarily focuses on one-step manipulation tasks, sequential manipulations require coordinated motions of a UAM's floating base, the manipulator, and the object being manipulated, entailing a unified kinematics and dynamics model for motion planning under designated constraints. By leveraging a virtual kinematic chain (VKC)-based motion planning framework that consolidates components' kinematics into one chain, the sequential manipulation task of a UAM can be planned as a whole, yielding more coordinated motions. Integrating the kinematics and dynamics models with a hierarchical control framework, we demonstrate, for the first time, an over-actuated UAM achieves a series of new sequential manipulation capabilities in both simulation and experiment.
BibTeX
@misc{su2023sequential,
  author = {Yao Su and Jiarui Li and Ziyuan Jiao and Meng Wang and Chi Chu and Hang Li and Yixin Zhu and Hangxin Liu},
  title = {{Sequential manipulation planning for over-actuated unmanned aerial manipulators}},
  year = {2023},
  eprint = {2306.14105},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2306.14105},
}
Figure