Meeting sustainable development goals via robotics and autonomous systems

The Sustainable Development Goals (SDGs) are an internationally agreed plan of action for people, planet, and prosperity, encompassing 17 goals and 169 targets across various areas. Many targets are interconnected, offering co-benefits but also potential trade-offs. Achieving the SDGs necessitates diverse investments and collaborative actions from governments, civil society, and the private sector. Progress towards the 2030 deadline remains insufficient, with ongoing challenges in areas like inequality, hunger, and carbon emissions. The COVID-19 pandemic further hampered progress, exacerbating existing issues. Technological advancements, particularly in robotics and autonomous systems (RAS), are transforming economies and societal interactions. RAS, capable of sensing, analyzing, interacting with, and manipulating their environment with minimal human intervention, are projected for widespread adoption. While the potential of digital technologies like RAS to facilitate SDG achievement is recognized—for example, AI's potential to contribute to 134 SDG targets—available information often focuses on individual SDGs rather than their holistic impact. Positive impacts are noted in healthcare, agriculture, and biodiversity conservation, but concerns exist regarding job displacement, pollution, and potential negative environmental impacts. A systematic understanding of RAS's overall impact on the SDGs is lacking, and current SDG plans rarely incorporate the potential of RAS, while RAS development often overlooks SDG considerations.

Existing literature highlights the potential benefits of RAS in specific SDG contexts. Studies show positive impacts on healthcare through enhanced surgical procedures and nursing care, on agriculture through improved weed control, and on biodiversity conservation through invasive species management. However, concerns are also raised about potential negative impacts on job markets, pollution and waste generation, biodiversity loss (e.g., through replacement of pollinators), and increased carbon emissions from transportation. This literature largely lacks a systematic evaluation of RAS's overall impact across all SDGs, highlighting the need for a comprehensive horizon scan.

This study employed a three-step horizon scan process to evaluate the opportunities and threats associated with RAS deployment concerning all SDGs. **Step 1: Online Questionnaire:** A structured online questionnaire was distributed to 102 experts from 23 countries, representing a diverse range of expertise and geographic locations. Participants evaluated the positive and negative impacts of RAS on up to three SDGs aligning with their expertise, using a 5-point Likert scale for each SDG target. In-depth explanations were sought for each response. Participants also assessed the overall impact of RAS on each SDG and the difficulty of predicting that impact. **Step 2: Group Discussions:** 44 participants were grouped based on their expertise, and each group synthesised the questionnaire responses for their assigned SDG. Each group identified the main positive and negative impacts, the three targets with the most significant positive and negative impacts, and an overall impact assessment. **Step 3: Workshop:** A workshop brought together representatives from each group to discuss co-benefits and trade-offs of RAS implementation across different SDGs. Data analysis involved content analysis of qualitative data from the questionnaire, group synthesis, and workshop discussions to identify key opportunities and threats, and quantify the salience of each opportunity and threat. Likert scale responses were used to visualize the overall impact and difficulty in predicting the impact of RAS on each SDG.

The horizon scan identified five key opportunities and four key threats associated with RAS deployment for achieving the SDGs: **Key Opportunities:** 1. **Replacing Human Activities:** RAS can automate unsafe, repetitive, or labor-intensive tasks in various sectors, improving efficiency and productivity. Examples include crop production, livestock management, waste management, and infrastructure maintenance. 2. **Supporting Human Activities:** RAS can assist humans in various tasks, especially in areas with labor shortages, such as elderly care and healthcare. RAS can enhance surgical procedures, patient movement in healthcare facilities, and facilitate health screenings. 3. **Fostering Innovation:** RAS can accelerate research and development across sectors, particularly in drug/vaccine development and renewable energies. RAS-led entrepreneurship can also stimulate job creation. 4. **Enhancing Access:** Autonomous transportation systems can facilitate access to remote or dangerous areas, improving disaster relief, service delivery, environmental conservation, and accessibility in general. 5. **Improving Monitoring for Decision-Making:** Automated data collection facilitates faster, more accurate, and more transparent decision-making across various sectors (infrastructure, resource distribution, wildlife monitoring, water quality monitoring). **Key Threats:** 1. **Reinforcing Inequalities:** RAS deployment could exacerbate existing inequalities due to affordability issues, job market transformations (reducing the demand for low-skilled labor), cultural contexts, and negative perceptions of RAS. Algorithmic biases in AI systems could amplify pre-existing inequalities. 2. **Negative Environmental Impact:** The environmental footprint of RAS, including resource extraction, energy consumption, pollution, and the potential for landscape simplification, poses a significant threat. Impacts on biodiversity, particularly for poorly known species in inaccessible habitats, are of particular concern. 3. **Resource Diversion from Tried-and-Tested Solutions:** Investments in RAS may divert resources from proven, less technologically intensive approaches to achieving SDGs, potentially hindering progress in socio-political domains. 4. **Inadequate Governance:** Lack of robust regulatory frameworks for RAS use and data ownership raises ethical concerns, increasing the risks of reinforcing inequalities and damaging the environment. Issues of data privacy, ownership of human behavioral data, and intellectual property rights are particularly important.

The findings indicate that while RAS deployment offers significant opportunities for achieving the SDGs, careful consideration of potential threats is essential. The overwhelmingly positive perceived net impact of RAS on the SDGs doesn't negate the identified threats; rather, it highlights the potential for substantial gains if risks are proactively mitigated. The difficulty in predicting the impact emphasizes the need for interdisciplinary collaboration between engineers, natural scientists, and social scientists to ensure a comprehensive understanding of RAS's societal and environmental implications. The observed lack of interaction between experts in RAS and those focused on SDGs related to poverty, equality, justice, and governance highlights the need for improved communication and collaboration.

RAS will fundamentally alter our interactions with technology and the environment. While offering substantial benefits for achieving the SDGs, realizing these benefits while minimizing unintended consequences requires proactive measures. A declared commitment to contribute positively to all SDGs during RAS design and deployment is crucial. Early stakeholder collaboration, enhanced dialogue, and increased engagement between engineers and sustainable development professionals are essential to harness the opportunities and mitigate the threats associated with RAS. Addressing identified threats (e.g., improving education, strengthening institutions) can also contribute to achieving the SDGs. The inclusion of RAS considerations in future SDG iterations is paramount to prevent unintended negative consequences and ensure that RAS contributes to a more sustainable and equitable future.

The study relies on expert opinions, which might be subject to biases and uncertainties. The horizon scan methodology focuses on future possibilities, and actual impacts of RAS may differ. The limited representation of low-and middle-income countries in some aspects of the study could affect the generalizability of findings to those contexts.