Robot Swarm Control Based on Smoothed Particle Hydrodynamics for Obstacle-Unaware Navigation

TL;DR

This work tackles obstacle-unaware navigation for robot swarms with limited sensing by introducing an SPH-based controller augmented with an indirect obstacle detector that infers collisions from velocity discrepancies. The method models the swarm as a fluid and adds collision-point repulsion to achieve robust obstacle avoidance without explicit obstacle sensing, achieving superior navigation and pattern formation across simulated and real environments. Key contributions include a practical indirect collision-detection mechanism, an SPH-based control framework with obstacle avoidance, and demonstrated real-time feasibility for swarms of up to around 100 robots. The approach promises greater autonomy and robustness for swarm robotics in unknown or sensor-limited settings, with potential applications in dynamic formation control and user-interface driven swarm tasks.

Abstract

Robot swarms hold immense potential for performing complex tasks far beyond the capabilities of individual robots. However, the challenge in unleashing this potential is the robots' limited sensory capabilities, which hinder their ability to detect and adapt to unknown obstacles in real-time. To overcome this limitation, we introduce a novel robot swarm control method with an indirect obstacle detector using a smoothed particle hydrodynamics (SPH) model. The indirect obstacle detector can predict the collision with an obstacle and its collision point solely from the robot's velocity information. This approach enables the swarm to effectively and accurately navigate environments without the need for explicit obstacle detection, significantly enhancing their operational robustness and efficiency. Our method's superiority is quantitatively validated through a comparative analysis, showcasing its significant navigation and pattern formation improvements under obstacle-unaware conditions.

Figures (6)

• Figure 1: Conceptual image of the situation we are challenged to solve and our proposed method. When robots collide with an undetected obstacle, they detect the collision indirectly from the change in their velocity. We add an element of repulsion force from these collision points to the SPH-based controller to achieve obstacle-unaware navigation.
• Figure 2: Appearance of the four simulation field environments: (a) Entry, (b) Dense pillar, (c) Barricade, (d) Pocket maze. The robot swarm (red dots) is initially located within the lime green square on the left side of the field. The robots run to the goal point, shown as a lime green dot on the right side, avoiding obstacles shown in black.
• Figure 3: Appearance of the robots and the experimental environment; (a) the eight swarm robots used in the experiment, (b, c) the environment imitating (b) the Dense pillar, and (c) the Pocket maze in the simulation.
• Figure 4: Trajectories of robot swarms in the simulation experiment. The blue lines in the figures represent the running trajectory of each robot, and the red dots represent the positions of the robots at the end of the simulation. The magenta dots represent the detected collision positions using our proposed method.
• Figure 5: Trajectories of the robot swarm with the proposed controller in the real environments. The content represented by colored lines and dots is the same as in Fig. \ref{['fig:simulation_results']}.