Nowadays,　wireless　sensor　networks　have　played　an　important　role　in　many　applications.　In　these　applications,　a　large　number　of　wireless　sensors　are　deployed　in　an　environment　to　collect　information　and　form　network　to　transmit　collected　information　to　the　base　station　or　sink.　Because　wireless　sensors　have　to　work　for　a　long　period　of　time　without　maintenance,　the　energy　consumption　of　wireless　sensors　has　a　great　impact　on　the　lifetime　of　wireless　sensor　network.　To　reduce　and　balance　the　energy　consumption　of　wireless　sensors,　many　information　transmission　routing　approaches　have　been　proposed.　However,　most　of　them　do　not　consider　all　energy　consumption　factors　of　wireless　sensors.　To　this　end,　an　innovative　information　transmission　routing　approach　based　on　Q-learning　is　proposed　in　this　paper,　which　enables　wireless　sensors　to　adaptively　select　suitable　neighboring　sensors　to　achieve　energy　efficient　information　transmission　in　a　decentralized　manner.　Based　on　the　proposed　approach,　a　wireless　sensor　first　collects　state　and　action　information　of　its　neighboring　sensors,　which　includes　information　transmission　direction　and　distance,　the　remaining　energy,　the　energy　consumption　for　information　transmission　and　information　transmission　action.　Then,　all　this　information　is　used　to　update　the　Q-values　of　neighboring　sensors,　so　as　to　enable　the　wireless　sensor　to　select　suitable　neighboring　sensor　to　transmit　information　according　to　Q-values.　From　simulation　experiments,　it　can　be　seen　that　the　proposed　approach　enables　wireless　sensors　to　reduce　and　balance　the　energy　consumption　of　wireless　sensors　and　extend　the　lifetime　of　the　entire　wireless　sensor　network.　IEEE