Embodied intelligence weaves a better future

Microrobots can interact intelligently with their environment and complete specific tasks by well-designed incorporation of responsive materials. Recent work demonstrates how swarms of microbots with specifically tuned surface chemistry can remove a hormone pollutant from a solution by coalescing it into a web.

An essential goal for the study of machine intelligence is the design of artificial agents that can fulfill specific missions efficiently with an integrated capability to perceive the environment and take actions accordingly. In the field of robotics, such intelligence is commonly embodied through sophisticated physical and algorithmic design, aiming to mimic the cognitive functions of natural living systems. When it comes to robots at micro- or nanometer scale, which hold promising potential in environmental and medical applications, the challenge lies in the miniaturization and assembly of functional hardware. Writing in Nature Machine Intelligence, Dekanovsky and colleagues address that challenge by designing a microrobot with chemically encoded intelligence1, for the efficient removal of hormonal pollutants.

The researchers describe the fabrication of a tubular microrobot with multiple functionalities (Fig. 1). The inner layer serves as the ‘engine’ of microrobot and is made of platinum. In the presence of hydrogen peroxide (H2O2) fuel, platinum catalyzes the decomposition of H2O2 into water and oxygen, and the subsequent ejection of oxygen bubbles from one side provides a powerful impetus to drive the microrobot through the liquid environments with dominant drag forces. The outer layer is a conductive polymer named polypyrrole (PPy) and is responsible for the chemical tasks, the condensation and decontamination of α-estradiol pollution (a model synthetic hormone). Magnetite (Fe3O4) nanoparticles are also incorporated into the PPy layer, which enables magnetic control over the motion and collection of the microrobots. The adsorption affinity of PPy to different molecules can be adjusted by chemically modulating the environments. For example, the surface charge of the microrobot gradually changes from a negative to a positive potential when the surrounding solution changes from basic to acidic, leading to an increasing attraction to negatively charged molecules. With its specifically designed surface chemistry, combined with a robust locomotive capability, the tubular microbot is able to effectively sort and concentrate α-estradiol on the microrobotic interface.

Fig. 1: Chemically programmable microrobots weaving a web from hormones.

The tubular microrobot, which consists of an inner layer of platinum (Pt) and an out layer of polypyrrole (PPy), can convert the substrate H2O2 fuel into self-propulsion impetus for the highly efficient removal of α-estradiol pollutant. Future research directions could involve the further development of actuation mechanisms (1), active and responsive capabilities (2), swarm intelligence (3) and modular systems (4), machine learning (5), as well as the integration of multifunctionalities including wireless communication (6), signal processing (7), autonomous therapy (8), and self-repairing (9).

It has no on-board sensors, actuators or microprocessors, yet can be said to show ‘intelligent’ behaviour in the sense of perceiving and interacting with the surrounding environment. Currently, it is a daunting task to endow microrobots with adequate computational capability to embody computer-based machine intelligence. Therefore, the development and incorporation of responsive materials such as in the work by Dekanovsky and colleagues play a vital role in making microrobots more intelligent. The strategy is not limited to chemically responsive microrobots, and it has also been demonstrated in many other types of systems, including those programmed by light, electric field, magnetic field, or a combination of them2. These advancements are leading to a myriad of intelligent microrobots for autonomous tasks.

An interesting aspect of Dekanovsky and colleagues’ work is that the microrobots not only respond to their chemical environments, but also locally interact with one another to weave macroscopic webs from the surface condensed estrogen fibers. The collective behaviour allows the transformation of α-estradiol molecules into one removable compact piece of fibrous texture and thus dramatically improves the decontamination efficiency, which differentiates this work from many other previous studies on the environmental applications of microrobots. Such swarm intelligence with group-level functionality is ubiquitous in nature, ranging from flocks of birds, schools of fish, to colonies of insects and bacteria, yet its involvement in microrobotic research is still in its infancy. In this regard, Dekanovsky and colleagues provide an inspiring example for the development of swarming microrobots.

Beyond their applications in environmental remediation, intelligent microrobots also hold great potential for revolutionizing biomedical fields (for example, targeted drug delivery and microsurgery), as the small size and robust mobility enable them to access hard-to-reach regions inside human body effectively and non-invasively3. Although the past decade has witnessed development in this field, considerable efforts will still be needed (see Fig. 1). First, more actuation mechanisms are needed. Compared to the use of toxic H2O2 fuel, microrobots would do better to directly consume the substance existing in surrounding environments for propulsion, which can avoid secondary pollution for environmental applications, and more importantly, is required for practical biomedical applications. The recent development of enzymatic microrobots emerges as a good choice4. Besides, biocompatible external power sources (for example, magnetic and acoustic fields) that can harmlessly penetrate living organs and tissues to actuate microrobots also provide a promising alternative5. Second, exploration of active and responsive material systems will be the key to strengthen the intelligence embodied in micromachines. For example, based on such materials, future microrobots may be able to sense the chemical signals associated with the target of interest (for example, oxygen gradient or change in pH value of a tumor), and autonomously perform locomotion steering, delivery and release of drugs6.

Third, given the limited capabilities of individuals, the implementation of swarm intelligence in artificial microrobotic systems is necessary. Achieving synchronous manipulation of a large group of robots with environmental adaptability7,8 will be an important challenge for improving task efficiency. And fourth, further efforts are needed to give microrobots multiple functionalities so they can complete a wide spectrum of tasks, such as wireless communication, signal processing, self-repairing, autonomous diagnosis and therapy. The recent surge of advanced micro- and nanofabrication technologies that succeed in the integration of circuit boards into micromachines9,10 has brought this goal closer. Collective robots with hierarchical functions or multicomponent modular systems11 can also contribute to the development of micromachines of increasingly complexity and functionality.

Dekanovsky and colleagues’ chemically intelligent microbots with collective behaviour, designed to fulfill a specific useful environmental task, are a stimulating example of the possibilities for advanced microrobotics.


  1. 1.

    Dekanovsky, L. et al. Nat. Mach. Intell. https://doi.org/10.1038/s42256-020-00248-0 (2020).

  2. 2.

    Palagi, S. & Fischer, P. Nat. Rev. Mater. 3, 113–124 (2018).

    Article  Google Scholar 

  3. 3.

    Nelson, B. J., Kaliakatsos, I. K. & Abbott, J. J. Ann. Rev. Biomed. Eng. 12, 55–85 (2010).

    Article  Google Scholar 

  4. 4.

    Tang, S. et al. Sci. Robot. 5, eaba6137 (2020).

    Article  Google Scholar 

  5. 5.

    Yan, X. et al. Sci. Robot. 2, eaaq1155 (2017).

    Article  Google Scholar 

  6. 6.

    Somasundar, A. et al. Nat. Nanotechnol. 14, 1129–1134 (2019).

    Article  Google Scholar 

  7. 7.

    Li, S. et al. Nature 567, 361–365 (2019).

    Article  Google Scholar 

  8. 8.

    Cichos, F., Gustavsson, K., Mehlig, B. & Volpe, G. Nat. Mach. Intell. 2, 94–103 (2020).

    Article  Google Scholar 

  9. 9.

    Bandari, V. K. et al. Nat. Electron. 3, 172–180 (2020).

    Article  Google Scholar 

  10. 10.

    Miskin, M. Z. et al. Nature 584, 557–561 (2020).

    Article  Google Scholar 

  11. 11.

    Alapan, Y., Yigit, B., Beker, O., Demirörs, A. F. & Sitti, M. Nat. Mater. 18, 1244–1251 (2019).

    Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Li Zhang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Jin, D., Zhang, L. Embodied intelligence weaves a better future. Nat Mach Intell 2, 663–664 (2020). https://doi.org/10.1038/s42256-020-00250-6

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing