This paper deals with the problem of adaptive energy-aware RREQ probability tuning in AODV reactive routing protocol. Two extensions of AODV are proposed and compared, namely: DFES AODV and FSARSA-AODV. First, DFES-AODV implements a fuzzy logic system with a time varying membership function for residual energy input. In comparison to SFES-AODV that is built on the topof a traditional fuzzy logic system, DFES-AODV is more energy efficient. Second, FSARSA-AODV uses a fuzzy extension with a criticonly architecture of SARSA RL algorithm. This hybrid algorithm overcomes the problem of empirical discretization of state space in SARSA-AODV protocol.