An empty space identification method, which is applied to a vehicle, and comprises: acquiring traffic events around a vehicle; establishing a Bayesian network according to a dependency relationship between the probability of occurrence of a traffic event and the probability of existence of an empty space; calculating the probability of existence of an empty space around the vehicle according to the traffic events and the Bayesian network; and if the probability of existence of an empty space is greater than a preset threshold, determining that an empty space exists. By establishing a dependency relationship between the probabilities of occurrence of different traffic events on the basis of the Bayesian network, and outputting the probability of existence of an available space in a range that the fields of view of sensors of the vehicle cannot reach, the accuracy of empty space identification during autonomous driving is improved.