Soft Robotics as an Enabling Technology for Agroforestry Practice and Research

Girish Chowdhary, Mattia Gazzola, Girish Krishnan, Chinmay Soman, Sarah Taylor LovellSustainability 2019 · DOI

TL;DR

This paper proposes soft robotics as a transformative technology to address the labor shortage and complexity challenges in sustainable agroforestry. By leveraging the unique properties of soft continuum arms, researchers are developing low-cost, dexterous robots capable of intricate tasks, paving the way for more profitable and environmentally beneficial agricultural systems.

What Problem Does This Paper Address?

The adoption of sustainable and diversified agroecosystems, particularly agroforestry, is hindered by a critical shortage of qualified human labor. These systems, while environmentally beneficial, demand more labor, skill, and knowledge than conventional annual monocultures. Traditional industrial robots, with their rigid and heavy arms, are ill-suited for the delicate and complex tasks required in these environments, such as precise harvesting of fruits and berries, weed removal, and pest detection among dense foliage. The high cost and limited adaptability of existing robotic solutions further exacerbate this problem, creating a significant barrier to entry for both small and large-scale farmers.

Methods

The core of this research lies in the development and integration of soft continuum robots with agricultural robots (agbots). These soft robots, inspired by natural structures like octopus tentacles and elephant trunks, utilize hyper-redundant, deformable elements. The paper highlights two key approaches for soft arm actuation:

  • Fiber-Reinforced Electromeric Enclosures (FREEs): These are hollow elastomeric membranes with embedded fiber families that allow for controlled extension, contraction, and rotation, enabling complex deformation patterns.
  • Asymmetric Continuum Arms: This approach combines different FREE configurations (e.g., extending and rotating FREEs) to achieve more complex, even spiral, motions for whole-arm manipulation.

The research also emphasizes the need for advanced dynamic modeling and simulation tools, specifically mentioning the Elastica solver, to overcome the trial-and-error nature of soft robot design. For autonomous navigation in challenging field environments, the paper discusses the use of LIDAR-based navigation to complement GPS, especially under dense canopies. Furthermore, Deep Reinforcement Learning (DRL) is proposed as a method for developing optimal control policies for these complex soft manipulators, learning from expert demonstrations and simulations.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Soft robotics offers a paradigm shift for agricultural robotics: Unlike rigid robots, soft robots can be manufactured at a lower cost, are lighter, and possess inherent compliance, making them safer and more adaptable for delicate agricultural tasks.
  • Agbots as collaborative teams: The vision is for teams of small, low-cost soft-armed agbots to work alongside humans, augmenting their capabilities and addressing labor shortages.
  • Morphological computation: The paper introduces the concept of "morphological computation" or "embodied cognition," where the physical design of the soft robot itself contributes to its control and task execution, reducing reliance on complex computational control.
  • The "phenotyping bottleneck" in agroforestry: The paper identifies a critical gap in the ability to collectively measure and analyze plant, soil, and microclimate data in field environments, hindering progress in plant breeding and sustainable agriculture research.
  • CROPS (Collective RObotic Observation and Phenotyping System): This conceptual integrated instrument, leveraging multiple connected sensors and autonomous robots, aims to address the phenotyping bottleneck by enabling scalable, reconfigurable, and autonomous data collection.

Key Results

While this is a concept paper, it presents preliminary work and promising directions:

  • Preliminary agbot design: A preliminary design for a soft-armed agbot is presented, featuring a continuum arm encased in a rigid shell for enhanced reach and a combination of rigid and soft embodiments.
  • Soft gripper development: A novel monolithic compliant gripper capable of simultaneously gripping and severing fruit by twisting has been designed and tested on larger fruits, with plans for miniaturization for berry harvesting.
  • Reinforcement Learning for control: Preliminary results show promising performance of Reinforcement Learning (specifically Deep Q-Learning) for position control of a soft continuum arm (BR2), demonstrating adaptive step sizes and robustness to tip loads. The learned policy was able to achieve an average error of 3 cm for 10,000 trajectories with a maximum tip load of 18g.
  • TerraSentia agbot capabilities: The TerraSentia robot is highlighted as a robust, ultra-compact, and low-cost autonomous platform suitable for carrying soft robotic arms and equipped with advanced machine vision for tasks like plant counting and width estimation.

Limitations and Future Work

The authors acknowledge several limitations and outline future research directions:

  • Challenges in soft robot design and control: Current soft robotic arms are difficult to design and control, slow to actuate, and have limited payloads. Significant advances in closed-loop control and sensing are needed for wide-scale adoption.
  • Speed and precision optimization: While low cost is a major advantage, the speed of operation and precision of soft robots will require optimization.
  • Integration of sensors and autonomy: Combining visual and LIDAR data for more robust navigation and developing sophisticated machine vision capabilities for tasks like identifying ripe berries, detecting diseases, and monitoring plant growth are crucial.
  • Real-world DRL training: The need for large amounts of data for DRL training presents a challenge for physical robots, necessitating the development of high-fidelity simulations and efficient learning strategies.
  • Extending to more complex manipulators: Future work includes extending the system for more complex manipulators with greater degrees of freedom and incorporating real-time feedback for closed-loop control and online policy adaptation.
  • CROPS development: The development of the CROPS instrument requires expertise in robotics, sensors, algorithms, wireless architecture, and field experiment design to create a truly integrated and autonomous data collection system.

Summary updated

Abstract

The shortage of qualified human labor is a key challenge facing farmers, limiting profit margins and preventing the adoption of sustainable and diversified agroecosystems, such as agroforestry. New technologies in robotics could offer a solution to such limitations. Advances in soft arms and manipulators can enable agricultural robots that can have better reach and dexterity around plants than traditional robots equipped with hard industrial robotic arms. Soft robotic arms and manipulators can be far less expensive to manufacture and significantly lighter than their hard counterparts. Furthermore, they can be simpler to design and manufacture since they rely on fluidic pressurization as the primary mechanisms of operation. However, current soft robotic arms are difficult to design and control, slow to actuate, and have limited payloads. In this paper, we discuss the benefits and challenges of soft robotics technology and what it could mean for sustainable agriculture and agroforestry.
BibTeX
@inproceedings{chowdhary2019soft,
  author = {Girish Chowdhary and Mattia Gazzola and Girish Krishnan and Chinmay Soman and Sarah Taylor Lovell},
  title = {{Soft Robotics as an Enabling Technology for Agroforestry Practice and Research}},
  booktitle = {Sustainability},
  year = {2019},
  doi = {10.3390/su11236751},
}
Figure