Self-organization and distributed cooperation of the unmanned swarm play an important role in completing the task of entrapping multiple targets. However, the cooperative entrapping of the unmanned swarm faces three challenges of limited communication, large scale, and limited location. Inspired by the cell swarm, we solve this problem based on the concentration diffusion model, including a local communication algorithm under the condition of limited bandwidth, a motion control algorithm under the condition of limited motion dimension, and a distance-based self-positioning method. Using the real robots named kilobots that is widely used in the research field as the experimental platform, we carry out experiments including static target entrapping, dynamic target entrapping, and large-scale swarm entrapping. Through statistical analysis, the feasibility of our method in completing these three tasks is verified.