A method and system for diagnosing autonomous vehicles is disclosed. The method includes identifying anomaly in a set of navigation parameters from a plurality of navigation parameters associated with the autonomous vehicle, such that each of the set of navigation parameters is above an associated risk threshold. The method further includes validating anomaly identified for each of the set of navigation parameters. The method includes generating a set of quality parameters for the set of navigation parameters in response to validating anomaly in the set of navigation parameters. The method further includes generating values of at least one motion parameter associated with the autonomous vehicle based on values of each of the set of quality parameters fed into a trained Artificial Intelligence (AI) model. The method includes controlling the at least one motion parameter based on the generated values.