What is the Net Climate Impact of 1 Billion Humanoid Robots?

“Our goal is to train a world model to operate humanoid robots at the billion-unit level.” - Brett Adcock, founder of Figure (AI Robotics)

Figure has just raised $675M in Series B to commercialize humanoid robots, on top of the $170M already raised. The 2-year-old startup has also secured a partnership with OpenAI and is now valued at a staggering $2.6B. The 2 trillion dollar company NVIDIA has just launched Projet GR00T to build foundation models for general humanoid robots. Meanwhile, Tesla, Apptronik, 1X, Agility Robotics, UBTech Robotics, and others are pressing ahead with their humanoids. China, a global heavyweight in industrial robotics, is actively pursuing a national strategy to accelerate the development of humanoids.?

The prospect of advanced humanoid robots becoming mainstream is exciting to many, and understandably so. Phasing out legacy machinery in favor of humanoids may enable some industries to become radically more efficient and decarbonize. Humanoid may even have a role to play in tackling the climate crisis, as discussed here. On the other hand, what will it take to manufacture, deploy, and operate humanoids at scale in terms of energy use, materials extraction, water use, and carbon emissions? What is the net climate impact of 1 Billion Humanoid Robots?

I’ve asked around, and it seems no one is quite sure what the answer is. “But you are raising an interesting question here!” a leading robotics expert replied. Will the positive impact on the climate outweigh the negative effects? How might we begin to explore this question?

The climate impact of humanoid robots will obviously vary based on their application and design. They may significantly reduce carbon footprints by optimizing resource utilization and scaling by scaling the deployment of climate solutions. They may contribute to lowering energy consumption, waste generation, and emissions in manufacturing processes. Additionally, humanoids may improve recycling processes, reducing CO2 emissions and material waste by automating waste treatment and recycling operations.?

On the other hand, advanced humanoid robots will necessarily use “Large World Models,” which process massive amounts of real-world data from various sources, such as IoT devices, sensors, cameras, and more, to comprehend and interact with the world in a way that mirrors human perception and cognition. Brett Adcock, founder of Figure, intends to train a world model to operate humanoid robots at the billion-unit level. If they’re anything like Large Langauge Models, then training and using Large World Models at scale will have a significant climate and environmental impact due to their staggering energy consumption, water use, and corresponding carbon emissions. Figure’s Humanoid uses both large world and language models.

Meanwhile, humanoid robots are made from materials like metals, plastics, and composites such as carbon fiber-reinforced polymers (CFRPs), kevlar, and laminates. According to estimates, the total weight of materials used in constructing a humanoid robot can range from a few kilos to tens of kilos, depending on the robot's size and intended use. These materials don’t “grow on trees,” nor will they come out of thin air. They need to be extracted, transported, and processed.?

Humanoids typically use a variety of batteries depending on their specific requirements. These include Lithium-Ion Batteries (Li-Ion), Nickel-Metal Hydride Batteries (Ni-MH), Lead-Acid/Sealed Lead Acid (SLA) Batteries, and Lithium Polymer Batteries (Li-Po). These batteries use critical minerals (rare earth elements) such as lithium, cobalt, nickel, manganese, and graphite. The demand for these minerals is expected to increase significantly to meet the growing needs of electric vehicles, battery storage systems, and humanoids.?

The extraction of these minerals is associated with a wide range of environmental problems, including habitat destruction, water pollution, and greenhouse gas emissions. Mining activities can lead to deforestation, soil erosion, and disruption of ecosystems, impacting biodiversity and local communities. Additionally, the processing of these minerals often involves the use of chemicals that can contaminate soil and water sources, posing risks to human health and the environment.?

Furthermore, the energy-intensive nature of mining operations contributes to greenhouse gas emissions, exacerbating climate change. The transportation of mined materials and the refining processes also add to the carbon footprint associated with critical mineral extraction. The concentration of mining activities in specific regions can lead to land degradation and resource depletion, affecting the long-term sustainability of these areas. Human rights abuses and forced labor are also intertwined with the extraction of rare earth elements. Lastly, we already know what happens when these batteries end up in landfills; they release toxins like heavy metals into the soil and groundwater.?

Scientists and sustainability experts will need to carry out rigorous and independent assessments better to understand the net impact of humanoids on the climate. They will need data from humanoid manufacturers. As we’ve seen from the limited studies that exist in relation to the environmental impact of aerial robotics (drones), robotics manufacturers are highly unlikely to share relevant data for the purposes of public, peer-reviewed studies. What’s more, the vast majority of drone manufacturers have shown zero interest in measuring their carbon footprints or in the need to become more sustainable. This mindset may well appeal to humanoid manufacturers.?

We need enlightened leadership and bold policies to ensure the regulatory and accountability frameworks are in place earlier rather than later to proactively manage the potential negative impact of humanoids, not just on the environment and climate but also on employment. I see more technologists pushing back against the fear that robots will create unemployment. They talk of all the beautiful things that make us human (like feelings, dreams, etc.), which robots can't replace. Tell that to everyone who has lost a job because they've been replaced by technology. Do you think it'll make them feel better??

In conclusion, listed below are some of the key steps that need to be taken to understand better the net climate impact of 1 Billion humanoids. What other steps should be added?

1. Energy Consumption: Measure the energy consumed during operation and standby modes to understand the robot’s overall energy usage. Likewise, energy consumption during the manufacturing process should be measured.
2. Material Usage: Evaluate the resources used in manufacturing the robot to assess its environmental footprint, including their extraction and transportation.
3. Emissions: Monitor emissions generated during production, operation, and disposal phases to quantify direct and indirect emissions (ideally Scope 1, 2, and 3).
4. Waste Generation: Analyze waste produced during manufacturing, maintenance, and end-of-life stages to gauge environmental impact.?
5. Resource Efficiency: Assess how efficiently the robot utilizes resources to optimize resource consumption.?
6. Life Cycle Assessments: Conduct a comprehensive life-cycle analysis to understand the environmental impacts at each stage of the humanoid’s life.
7. End-of-Life Management: Develop strategies for responsible disposal or recycling of humanoids to minimize environmental harm.
8. Sustainability Practices: Implement sustainable humanoid design and operation practices to reduce environmental consequences.?
9. Data Collection and Reporting: Collect data on all environmental aspects, analyze findings, and report on the humanoids’ environmental performance.?