Can we detect aurora in exoplanets orbiting M dwarfs?

New instruments and telescopes, such as SPIRou, CARMENES and TESS, will increase manyfold the number of known planets orbiting M dwarfs. To guide future radio observations, we estimate radio emission from known M-dwarf planets using the empirical radiometric prescription derived in the solar system, in which radio emission is powered by the wind of the host star. Using solar-like wind models, we find that the most promising exoplanets for radio detections are GJ 674 b and Proxima b, followed by YZ Cet b, GJ 1214 b, GJ 436 b. These are the systems that are the closest to us (<10 pc). However, we also show that our radio fluxes are very sensitive to the unknown properties of winds of M dwarfs. So, which types of winds would generate detectable radio emission? In a reverse engineering calculation, we show that winds with mass-loss rates dot{M} > kappa_sw /u_sw^3 would drive planetary radio emission detectable with present-day instruments, where u_{sw} is the local stellar wind velocity and kappa_sw is a constant that depends on the size of the planet, distance and orbital radius. Using observationally-constrained properties of the quiescent winds of GJ 436 and Proxima Cen, we conclude that it is unlikely that GJ 436 b and Proxima b would be detectable with present-day radio instruments, unless the host stars generate episodic coronal mass ejections. GJ 674 b, GJ 876 b and YZ Cet b could present good prospects for radio detection, provided that their host-stars winds have dot{M} u_sw^3 > 1.8e-4 Msun/yr (km/s)^3.