The paper presents the design, simulation and evaluation of an improved obstacle avoidance model for the lawnmower. Studies has shown that there has been few or no work done has on the total minimization of computational time in obstacle avoidances of land mower. Sample image data were collected through a digital camera of high resolution. The obstacle avoidance model was designed using the Unified Modelling Language tools to model the autonomous system from the higher-level perspective of the structural composition of the system. Automata theory was used to model two major components of the system, which are the conversion of the colour image to binary and the obstacle recognizer components by neural network. The model was simulated using the and evaluated using the false acceptance rate and false rejection rate as performance metrics. Results showed that the model obtained False Acceptance Rate and False Rejection Rate values of 0.075 and 0.05 respectively. In addition, the efficiency of the proposed algorithm used in the present work shows that the time taken to avoid obstacles was faster when compared with another existing model.