A neuro-fuzzy system is developed for behavioral based control for reactive navigation. Our goal is to develop a robot which works in room like environment. The start position and end position is given to it. It gets its distance from the obstacles by range sensors and traces its rout to the end target by reactive navigation. The inputs to the system consist of (i) range information acquired by sensors and (ii) the heading angle (angle between the robot’s heading direction and the target direction). The Neural network at input transforms this input to a reference motion direction and the fuzzy system uses this along with a modified form of the sensor’s input to yield wheel velocities. We have taken the neuro- fuzzy system developed by Joshi M.M. [4] as reference. This system is conceptually analogous to that indicated by Wei. Li [14]. In [4] they have concluded that for behavior based reactive navigation the Neuro-fuzzy system perform better than Neuro, neuro-neuro and fuzzy system. The actual simulation environment is different than [4]; we have developed the algorithm in Turbo C language. In our case we have developed the algorithm for neural networks and fuzzy logic, In Matlab which are available as functions. The environment can take any shape of obstacles as input. The heading angle and apparent angle correction methods are also different. These neuro- fuzzy based reactive methods are also adequate for moving obstacles avoidance. We have tried to develop actual robot based on neuro- fuzzy system. We have started with using AVR controller. We have tried various c-compilers for AVRs. We have realized that the computational complexities required for Neuro-fuzzy system is very high. Therefore 8-bit MCU (microcontroller unit) will not adequate. We require MCU with 32- bit ALU and with saturation mathematics. The TI DSPs of TMS320c2000 or TMS320c54X or ARM7TDMI will be adequate. We have found the distance sensors and their supplier. We have developed the mechanical assembly for the robot using stepper motor. We will realize the problem associated with hardware development using the fuzzy logic given velocities.