The multisensor fire-detection algorithm is one of the current important issues in the field of fire-detection systems for intelligent buildings. This paper proposes an adaptive fusion algorithm for fire detection, and uses a smoke sensor, flame sensor, and temperature sensor to detect fire incident. In reality, the phenomenon of the fire incident may have smoke, flame, and high-temperature situations. However, these signals may happen simultaneously or sequentially. We use adaptive fusion algorithms to a more reliable decision. However, the adaptive fusion algorithm is more complex in real conditions. Therefore, we use a Taylor expression to modify the adaptive fusion algorithm and simulate to compare with results on first-order, second-order, and third-order expressions. The modified adaptive fusion method can provide adequate reliable fusion for fire detection. We use computer simulation to improve the adaptive fusion algorithm that is accurate and adequate. Then, we design a fire-detection module using an ionization smoke sensor (TG-135), temperature semiconductor sensor (AD590), and ultraviolet sensor (R2868). The experimental results of the fire-detection module demonstrate that it can detect fire incidents in a variety of conditions. Finally, we implement the real-time fire-detection module in an intelligent security robot (Chung Cheng I). If a fire incident occurs, the security robot can find the fire source using the fire-detection module and transmit the detected message to the user via the Internet and a GSM modem.