A new navigation method for mobile robot based on WSNs, including whole network navigation path planning and local node approaching algorithms, is proposed to provide navigation path with minimal distance and lowest cost in location-free Wireless Sensor Networks(WSNs). The method constructs cost function by combining sensor data, and builds a pseudo-gradient potential field to plan the best path for mobile robot. Mobile robot approaches the nodes on the path individually by detecting Receiving Signal Strength Indicator(RSSI). Simulations show that the method can guide mobile robot to the target node rapidly according to its navigation requirements.