Traditional Fire Detectors respond mostly to Smoke and Heat. The smoke sensors function by detecting the presence of smoke particles either in a photoelectric chamber or in an ionization chamber. These sensors being placed on the ceiling Fire is seldom detected in the incipient stage. Loss due to fire damage not only accounts for equipment and property destruction, but also loss of data, interruption of service, cleanup and recovery cost. On an average about $250,000 per incident. The key to control these damages are not only to detect fire as early as possible but also identify exactly the origin of incident. All over the world the latest technologies are being tested upon and implemented to detect the fire at early stage. However these require expensive proprietary solutions and may not be easily deployable in existing infrastructure. Technologies do exist today to detect fire at an early stage but are expensive in nature and requires pre-engineered planed deployment. This paper demonstrates means to detect the fire instantly and extinguished at initial stage. An autonomous robot equipped with advanced fire detection technology detect fire at initial stage, extinguished by small conceived extinguisher, sound hooter and also send message to pre assigned number through GSM modem. A novel approach using color sensor TCS3200 and simple LDR (Light Dependent Register) makes the system highly cost effective. It is effectively a fire surveillance system that continuously read sensor values and received data are processed by various complex algorithms to ensure fire detection with highly reduced false alarm and immediate action. It also covers a large area and thus the system costs are minimized. It is highly useful for domestics as well as industrial environment. The total system cost is less than $200. It is important to note that electrical supplies must be cut down in case of a fire incident and thus also takes the fire detection system offline. This system being self-sufficient and battery powered can still function.