In this study, we used Cricket equipments based on wireless sensor networks to study the pursuit-evasion strategy of multi-robots in a continuous map. In order to improve the efficiency, two new pursuit-evasion strategies were proposed. To meet the need of a continuous map, a greedy algorithm was utilized to deal with boundary problem when a pursuer or an evader was near the field boundary. The statistical results show the converging-attack pursuit strategy outperforms in pursuing the evader with different escaping strategies, such as randomly moving, active escaping strategy and vector-wards escaping strategy. The vector-wards escaping strategy is also effective in delaying the capture of evaders.