Sustainability Robotics
| Location | Dübendorf |
| Guest | Mirko Kovac |
| Interviewer | Riva Pinto |
How can robotics help protect our environment and maintain infrastructure sustainably? At the Laboratory of Sustainability Robotics at Empa and EPFL, Mirko Kovač and his team are developing drones and autonomous systems that can gather data, carry out repairs, and even biodegrade after use. In this interview, he explains how bio-inspired robotics can support both society and the environment.
Riva Pinto: At the Laboratory of Sustainability Robotics at EMPA, you and your team develop robots that collect environmental data and build structures in challenging terrains. What’s the core idea behind your work?
Mirko Kovač: Our mission is to explore how robotics and AI technologies can support the United Nations’ Sustainable Development Goals. We combine environmental science, material science, and robotics into one interdisciplinary field. That is why our lab is jointly hosted by Empa and EPFL. At EPFL we are part of ENAC, the School of Architecture, Civil and Environmental Engineering and at Empa we are at the department of Engineering Sciences and at Empa we are at the department of Engineering Sciences.
The idea is to develop robotic tools that collect high-quality data in more efficient and less invasive ways. Many current methods rely on manual interventions, which can be expensive, slow and disruptive. For example, collecting data in the Arctic or in forests often requires large teams and heavy equipment. This has a real environmental impact. Robotic systems can reduce cost and risk while delivering more precise and frequent data.
Pinto: What are some of the concrete systems you’ve developed?
Kovač: One example is the FireDrone. It is built with a new type of aerogel developed with Empa. This material is extremely heat resistant, lightweight and functionally suitable for the robot body. The FireDrone can fly in high-temperature environments and is useful for firefighting or industrial inspection. It is now becoming a startup with funding from Innosuisse and Venture Kick among others.
Another example involves aerial-aquatic drones. These can fly and dive into water to collect samples or environmental DNA. They are particularly valuable in regions like the Arctic, where data is scarce but urgently needed. These systems help monitor biodiversity and environmental change with low impact and cost.
We are also developing biodegradable drones, which we call transient robotics. These are made from cellulose and carbon-based inks. If lost in the field, they break down naturally instead of turning into e-waste. This approach makes it possible to deploy large numbers of small robots without adding to environmental problems.
A final example is our aerial 3D printing drones. These flying robots deposit material while in flight. Unlike traditional 3D printers, they are not constrained by the size and print envelope of the printer. They can build structures or perform automated repair at height or in remote or hard-to-reach areas. We have worked on this for over a decade and have several industry ready systems available. One of our projects was published and featured on the cover of Nature, and a new review paper just came out in Science Robotics. We see great potential in using these systems for construction or infrastructure repair.
Pinto: You mentioned aerial robots that can deposit materials and build structures. What role could autonomous construction play in the future of our built environment, especially in remote areas or disaster zones? And how does sustainability factor into this?
Kovač: Aerial additive manufacturing could be especially useful for repair. Take a bridge, for instance. If it is damaged, repairs often require shutting it down, building scaffolding, and sending people to work at height. This is expensive and dangerous.
Flying robots could carry out small or temporary repairs quickly and safely. They could seal cracks in dams, repair building facades, or maintain offshore wind turbines and pipelines. This extends the life of infrastructure with less cost and risk.
As infrastructure ages, the ability to maintain rather than replace becomes more important. Using autonomous systems for this purpose helps reduce environmental and financial impact.
Pinto: Your work often draws inspiration from biological systems. Can you discuss how nature informs the design and functionality of your robots?
Kovač: The term “bioinspired” is often used loosely. A robot with four legs is not necessarily truly inspired by nature. What matters is identifying functional principles from biology and applying them meaningfully.
We do not copy shapes directly. Instead, we try to abstract the underlying mechanism or behavior and then implement it using engineering tools. In traditional robotics, design follows a sequence. You simulate, build, then test. But the environments we work in, such as unstable terrain or turbulent air and water, are difficult to simulate accurately. This makes the standard approach less effective. In contrast, natural systems evolve with their environments. Body, control, sensors, and surroundings are all connected. Muscles, for example, develop through interaction with the world. We aim to bring this co-evolution of body and function into robotics.
To describe this, we introduced the idea of Physical Artificial Intelligence. It is a framework that brings together materials, structures, actuation, and control in one integrated, iterative design process. The goal is not just to make robots that look natural, but ones that behave in ways more similar to living organisms.
Pinto: How does this integrated approach relate to more data-driven forms of AI? Where do algorithms and learning systems fit into your robotics work?
Kovač: Intelligence in robotics often means processing sensor data to make decisions. AI tools like classification algorithms or large language models can analyze this data and guide the robot's actions. This allows us to move beyond rigid, model-based controllers, making robots more adaptable and robust.
But I believe algorithmic intelligence alone is not enough. We also need to think about intelligence in terms of form and materials. There is a concept called morphological computation. It means that a robot’s structure can contribute to its function. For instance, a soft, compliant robotic hand can conform to the shape of an object without computing every motion. Nature relies on this kind of intelligence constantly.
That is why we work with the concept of physical artificial intelligence. It brings together sensing, actuation, materials, and control in a single evolving process. We also see this as a future model for educating engineers to design more capable and integrated systems.
Pinto: Your robots are primarily developed for civil applications such as environmental monitoring or disaster response. But with growing global interest in robotics for defense, how do you see your work in that wider context? Do you see a risk that your technologies might be applied in unexpected settings?
Kovač: Robotics is a dual-use field by nature. Like many technologies, it can be used for both civilian and military purposes. That is simply a reality. However, military systems are typically very different in how they are designed and built. They need to function under specific constraints, often without access to civil infrastructure or standard sensors. Because of these needs, the technologies diverge in important ways. Our systems might resemble military ones at a glance, but the actual technology, design goals, and applications are often fundamentally different. It is important to keep that distinction clear.
Pinto: Looking 10 or 15 years ahead, what would you like to achieve with your work? And how do you imagine the coexistence of humans, the environment and autonomous systems in the future?
Kovač: I would like our robots to become mobile agents that support both ecosystems and human activity. I see them as artificial organisms that help us monitor and protect fragile environments such as the Arctic, mountain areas, or rainforests. In cities, they could also play a positive role.
The goal is to create systems that are non-invasive and well integrated. Today, much data collection relies on heavy equipment that disrupts nature. Imagine instead small, smart robots that can carry out precise tasks with minimal impact. This also applies to agriculture. Small robotic tools could reduce the need for large machines and chemicals, helping both food production and the environment.
The idea is not to replace human labor, but to complement it. Especially in dangerous tasks such as offshore maintenance, mining, or forestry, robots can improve safety and reduce risk. With thoughtful integration, these systems can become valuable partners in preserving the environment and improving safety for people.