Automated decision-making by chemical echolocation in active droplets

Abstract

Motile microorganisms, like bacteria and algae, unify abilities like self-propulsion, autonomous navigation, and decision-making on the micron scale. While recent breakthroughs have led to the creation of synthetic microswimmers and nanoagents that can also self-propel, they still lack the functionality and sophistication of their biological counterparts. This study pioneers a mechanism enabling synthetic agents to autonomously navigate and make decisions, allowing them to solve mazes and transport cargo through complex environments without requiring external cues or guidance. The mechanism exploits chemo-hydrodynamic signals, produced by agents like active droplets or colloids, to remotely sense and respond to their environment - similar to echolocation. Our research paves the way for endowing autonomous, motile synthetic agents with functionalities that have been so far exclusive to biological organisms.

Figures (4)

• Figure 1: Automated decision-making by chemical 'echolocation'. (a) Schematic showing the idea of chemical echolocation. The agent (grey sphere), e.g., a droplet swimmer or an autophoretic colloid, creates a chemical (or chemo-hydrodynamic) signal (background color) that is ‘echoed’ by dead ends in a maze and effectively repels the agent based on physical forces. The agent consistently makes correct navigational decisions (green ticks) toward the target (open end/reservoir) without requiring guidance from a source or external controls. (b) Simulated trajectory (white-green line, color indicates time $t$) showing navigation by the chemical echolocation strategy. (c) Simulated trajectory (white-green line) of a source-seeking agent solving a maze by sensing a chemical gradient created by an external chemical source (red dot) at the exit. Background colors in (b, c) show the normalized chemical concentration $c/c_{\text{max}}$. (d) Histogram of exit times for random initial conditions near 'start' in (b) and (c). (e) Average exit time as a function of the distances to the 'exit' region in (b) and (c). The signal-to-noise ratio (SNR) is everywhere greater than 1 for the echolocation strategy but smaller than 1 for the source-seeking strategy at large distances from the exit, making it ineffective in large mazes. Parameters: $D^*=10^2$, $B^*=2\times 10^4$, $M^*=0$.
• Figure 2: Experimental realization of automated decision-making in droplet swimmers. (a) Schematic of the droplet showing the asymmetric distribution of the released PSS, leading to spontaneous symmetry breaking and self-propulsion. (b) Speed of the droplet in a rectangular channel. The inset shows the corresponding trajectory in this confined channel. (c) Schematic of the droplet motion in a large complex maze. (d) Variation of the droplet velocity at a junction. The inset shows the trajectory of the droplet. (e) Success fraction (SF) of the droplet at each junction on the first attempt versus distance from the exit. (f) Autonomous large cargo delivery without requiring external control.
• Figure 3: Comparison of experiments and simulations. Typical trajectory of a droplet in the maze from (a) experiments and (d) simulations, with time shown in colorbars. Histogram of exit times of droplet trajectories in (b) experiments and (e) simulations. The solid lines are Gaussian fits to the data. Mean exit time of the droplet from different distances to the exit point in (c) experiments and (f) simulations. Error bars in c are first-to-third quartile of the distributions. Parameters: $D^*=2\times 10^2$, $B^*=-8\times 10^4$, and $M^*=10^{-1}$ (for panels d,e,f).
• Figure 4: Robustness of chemical echolocation for varying maze geometry and parameters. Representative agent trajectories from (a,b) experiments and (c,d) simulations demonstrate successful navigation through complex mazes with extended dead-end lengths and multiple branches, highlighting the effectiveness of the chemical-echolocation-based decision-making strategy. Parameters for panels (c,d) are the same as in Figs. \ref{['fig3']} (d-f). Background colors in (c, d) show the normalized chemical concentration $c/c_{\text{max}}$ when the droplet reaches the exit. (e) Exit times and success fraction of the droplet for the maze shown in Fig. \ref{['fig3']}d for different values of the reduced chemotactic sensitivity $B^*$. The black solid lines denote moving averages. The black dashed lines indicate the average values of the exit times (Fig. \ref{['fig3']}b) and success fractions (Fig. \ref{['fig2']}e) of the droplet from the experiments. The orange dashed line indicates the value of $B^*=-8\times 10^4$ used in the Figs. \ref{['fig3']} (d-f). Parameters: $D^*=2\times 10^2$ and $M^*=10^{-1}$.