An Open-source, Re-Configurable and Economical Robotic Manipulator Platform

While balancing NDAs and contractual obligations posed challenges, engaging with the open-source robotics community has profoundly influenced my approach to research and collaboration.

- Kartikeya Walia

Tell us a bit about your research.

My name is Kartikeya Walia and as a researcher with a background in mechanical engineering, I am particularly focused on creating accessible, sustainable, and impactful robotic solutions. My work combines industry collaboration, innovative design, and a commitment to fostering inclusivity in robotics to address real-world challenges.

Tell us about a project you were involved in which used Open Research practices and principles?

Robots are becoming increasingly prevalent across various industries, playing a crucial role in shaping the smart factories of the future, driven by Industry 4.0 principles. However, many industries face an immediate challenge in adopting robotics due to a lack of expertise, relying heavily on robotic integrators to suggest solutions. This dependence often leads to vendor lock-ins and further reliance on third parties for any necessary workflow modifications. Additionally, the high cost of robots discourages many small and medium-sized enterprises (SMEs) from exploring the potential benefits of automation.

My PhD, in collaboration with PepsiCo (collaborative partnerships), revolved around developing a robotic system to address the complexities of the current robotics integration paradigm and produce an open-source, easy to integrate modular robotic manipulator platform. CRISP (Customisable Robot with Inexpensive and Sustainable Performance)offers additional benefits including being re-configurable, economical and reusable making it a more sustainable automation solution.

Describe the open research practice(s) employed in your study. Why did you select them?

The research focused on ensuring the developed robotic system remained accessible and open-source throughout its lifecycle, as it was primarily developed for end users with low technical expertise. Key features like ease of assembly, integration, deployment, operation, repair, decommissioning, and reusability were identified to enhance the system’s accessibility.

Collaboration with PepsiCo included multiple design review sessions to refine designs with expert industry feedback. Play-days were conducted for PepsiCo and NTU Engineering teams to engage with prototypes, gathering valuable input (co-design and production). I also shared insights with PepsiCo through the NTU-PepsiCo framework agreement (Knowledge Exchange).

Utilising open-source platforms like ROS (robot-operating-system)¹ and GitHub² ensured future-proofing of the source code and leveraged community contributions. While research outcomes are exclusively accessible to PepsiCo in the Food and Beverage sector due to contractual constraints, PepsiCo supported dissemination through presentations at national and international conferences³ and publication in peer-reviewed, open-access journals^4,5 following their approval.

Did you face any challenges in the project, and how did you overcome them?

Transitioning from a background in Mechanical Engineering to pursuing a PhD in robotics presented numerous learning opportunities and challenges. A significant learning experience was implementing ROS to develop functional Proof-of-Concept demonstrators. Despite encountering multiple software development challenges and errors, the supportive open-source ROS community⁶ played a crucial role. Access to reusable libraries on GitHub and communication via the ‘Issues’ section facilitated learning and troubleshooting, minimising the need to reinvent existing solutions. This greatly aided my development process and motivated me to share my work openly, ensuring I give back to the robotics community.

Using ROS ensured seamless future implementation and software modularity, while GitHub facilitated version control and file sharing^7,8 (Open Data).

A key consideration throughout the project was ensuring a lower system complexity. This was evaluated using a developed 'Assembly-Complexity-Index' (ACI) scale. The volunteers from PepsiCo and NTU Engineering followed a comprehensive Assembly Manual⁸ and provided valuable feedback for design iterations and improvement. The ACI framework quantitatively assessed workload and complexity using NASA Task-Load-Index (NASA-TLX) and Task-Complexity-Index (TCI) scales. This framework, pivotal in objectively measuring system complexity, will be presented at the International Applied Human Factors and Ergonomics Conference 2024⁹, with proceedings published in an open-access article.

A significant challenge was balancing openness and contractual limitations. As the project was conducted under an NDA with PepsiCo, all outcomes related to the Food and Beverage sector were restricted from public dissemination. To address these limitations, I worked closely with PepsiCo teams to establish clear boundaries for what could be shared openly. For instance, while technical design principles and source code were made publicly available, design details, design files, general assembly procedures, and sector-specific implementations remained confidential. Additionally, in collaboration with the IP and commercialization team at NTU, several features of the design were protected by filing for a patent (UK Patent Application No. GB2312821.8, currently pending). This careful delineation ensured compliance with contractual obligations while still contributing meaningfully to the open-source community.

What has been the impact of adopting open research practice(s) in your project?

Open research practices significantly shaped my project, emphasising usability and openness in the development of CRISP. These advancements focussed on ensuring reliability, whilst accommodating low-technical end-user expertise, thereby enhancing inclusivity in robotics. Following PepsiCo’s approval, the research was disseminated through open-access publications, conference presentations, and accessible files for public use.  While core technical components of CRISP, including principles like generative design, solutions to geometric design problem (GDP), inverse kinematics, general dynamics solutions for modular manipular robotics, and the source code, were made publicly available, certain proprietary aspects remained confidential. Specifically, hardware design details, design files, general assembly procedures, and sector-specific implementations were withheld due to their sensitive and application-specific nature.   This selective release approach ensured that SMEs and the broader robotics community could benefit from the project while safeguarding PepsiCo’s competitive edge in the Food and Beverage sector. By maintaining transparency in this process, trust was fostered with collaborators, and the spirit of open research was upheld.

Additionally, completing the ‘2022-IEEE RAS-Seasonal-School on Reproducible Research, Performance Evaluation, and Benchmarking in Robotics’¹⁰ equipped me with best practices for research reproducibility and performance evaluation in robotics. Embracing open practices has enabled me to receive significant recognition and foster valuable networking and collaboration opportunities whilst sharing the knowledge. Notably, my article in the Q1 journal ‘Polymers’⁵ received the highest citations among my publications. The CRISP project received ‘high commendation’ at the Robotics-and-Automation-Awards 2023¹¹ in the ‘Sustainability’ category. I also secured Early-Career-Researcher-Placement funding from the EPSRC Connected-Everything Network with the outcome published in a newsletter^12,13 for the wider audience.

What did you learn from making this project ‘open’? Do you have any advice for others considering adopting open research practices?

Embracing openness in research has been transformative. Working within an open environment has fostered invaluable collaborations, enhanced reproducibility, and accelerated innovation. While balancing NDAs and contractual obligations posed challenges, engaging with the open-source robotics community has profoundly influenced my approach to research and collaboration.

If I were to approach this project again without commercial or contractual limitations, I would prioritise making the manipulator design files public. This would include ensuring parts could be easily sourced, reducing hardware costs, and redesigning robotic joints to work with more affordable electronics. Although this could introduce reliability concerns, it would offer an excellent opportunity for others to learn by recreating the system, encouraging widespread engagement and knowledge dissemination.

For early-career researchers, my advice is to proactively negotiate the boundaries of openness in collaborative projects. Strive to align stakeholder expectations while adhering to open research principles wherever possible. Engaging with the open research community not only enhances the visibility and impact of your work but also contributes to collective growth and innovation in your field.

References

1. Stanford Artificial Intelligence Laboratory et al. (2018). Robotic Operating System. Retrieved from https://www.ros.org/
2. Github. (2020). GitHub. Retrieved from https://github.com/
3. Walia, K., Navaraj, W., Khalid, A., Khan, A., & Breedon, P. (2022). A low-cost reconfigurable industrial robot design utilising additively manufactured components. IEEE/RSJ International Conference on Intelligent Robots and Systems, Kyoto, Japan, 23-27 October 2022
4. Walia, K., Khan, A., & Breedon, P. (2021). The generative design process for robotic design applications. Journal of Additive Manufacturing Technologies, 1(2), 528-528.
5. Walia, K., Khan, A., & Breedon, P. (2021). Polymer-based additive manufacturing: process optimisation for low-cost industrial robotics manufacture. Polymers, 13(16), 2809.
6. ROS Discourse Community. Retrieved from https://discourse.ros.org/
7. Walia, K., CRISP source-code (Version 1.0.2) [Computer software] Available at https://github.com/KartikeyaWalia/CRISP_trials/tree/main
8. Walia, K. CRISP files and Assembly Manual. Available at https://github.com/KartikeyaWalia/CRISP_files
9. Applied Human Factors and Ergonomics International Conference, 2024, Hawaii. Retrieved from https://www.hawaii.ahfe.org/
10. IEEE Ras Seasonal School on reproducible research, performance evaluation and benchmarking in Robotics. IEEE Robotics and Automation Society. https://www.ieee-ras.org/about-ras/ras-calendar/event/2104-ieee-ras-seasonal-school-on-reproducible-research-performance-evaluation-and-benchmarking-in-robotics
11. “2023 Winners.” The Robotics & Automation Awards 2023, https://www.roboticsandautomationawards.co.uk/2023-winners
12. Connected Everything, Newsletter (2022) https://connectedeverything.ac.uk/wp-content/uploads/2022/07/newsletters_issue14.pdf (14) pp-7
13. Walia, K. Connected Everything, Early Career Researcher Placement, 2022 Retrieved from https://connectedeverything.ac.uk/wp-content/uploads/2022/07/connected-everything-placement_kartikeya-walia.pdf

Kartikeya Walia | Nottingham Trent University