An agent based system is used to find the air pollution in traffic area and generate recommendation to the road users. The air pollutants data are collected using static air pollution monitoring stations and mobile sensors. These mobile sensors are low cost. Using this data find Air Pollution Index (API) then creates a road network. This network is represented as a weighted graph. From this graph find the least polluted path. Dijkstra algorithm is used to find the least polluted path. To find the concentration of pollutants use Artificial Neural Network (ANN) is used. Here need to process large amount of data so Hadoop framework is used.