Programming swarm behaviors usually means dealing with multiple tools, simulators, and robot-specific APIs. PyBuzz aims to simplify this: - One programming interface for different robot types - Easy transition from simulation to real robots - Common swarm behaviors built-in (consensus, stigmergy, virtual physics) - Works with existing robotics tools (ROS, Buzz)
Benchmarking Localization Strategies for Swarm Robotics — Swarm Intelligence Journal
Robot swarms need to know where their neighbors are to work together. This is hard indoors or when GPS isn't available—there's no global reference, sensors are noisy, and communication is limited. We compared four common localization approaches: - Infrared (IR) range-and-bearing sensors - Visual-inertial odometry (VIO) with communication - Ultra-wideband (UWB) with fixed anchors - Camera-based spatial model (CoViS-Net) We tested these in large-scale simulations, photorealistic simulations, and real experiments with ground and aerial robots doing three different swarm behaviors. Main findings: - No single approach works best everywhere—each has trade-offs in accuracy, setup complexity, computation, and communication needs - Localization errors directly affect how well swarm behaviors work, especially those that depend on precise directions - The results provide guidance for choosing a localization method based on your environment and task - Tested across ARGoS simulations, Isaac Sim, and real robot experiments Explore this page to dive deeper into the workings of this model. For real robots doing pursuit, see the video below.
