That actually exists. It's called turbocompounding (https://www.dieselnet.com/tech/engine_whr_turbocompound.php). Sometimes the turbine is coupled directly to the crankshaft and there's no compressor. Sometimes there's no turbine (http://mechstuff.com/differences-between-superchargers-vs-turbochargers/).
Here's the power a turbine can produce:
dW/dt [W] = eff*dm/dt * c_p_mean * T_inlet*[(p_outlet/p_inlet)^((k-1)/k)-1]
T_inlet: Inlet temperature [K]. Usually 400-500 ºC for Otto engines and a around 600 ºC for Diesel engines. Let's take 873 K.
c_p_mean: Mean heat capacity at constant pressure [J/kgk]. Mean between T_inlet and T_outlet. For air it would probably be around 1050 J/kgK, but consult tables.
eff: Efficiency of the turbine. 0.80 is a typical value.
p_outlet: Outlet pressure [Pa]. Normally atmospheric pressure (101,325 Pa).
p_inlet: Inlet pressure [Pa]. Let's assume 200,000 Pa (2 bar), because the expansion isn't full.
k: Adiabatic index, usually 1.4 for air and similar for exhaust gases.
m: Mass [kg]. Its temporal derivative is mass flow [kg/s]. The same mass that enters must exit (ignoring leaks through the gaps). Let's say 4*10^-3 kg (I can't remember the figures I knew now). And 4 kg/s as the exhaust process takes place at a few miliseconds.
W: Work [J]. Its temporal derivate is power [W].
dW/dt = -514,458 W = -514.458 kW, which is too high. It could be from 1 to 10 kW, for example, depending chiefly on the mass flow and the expansion ratio (p_outlet/p_inlet).
Sorry for the terrible format, but I'm new to this forum and I don't know how to format yet. If you are going to convert the power to electrical power, multiply to 0.90 (usual efficiency of a electric generator/motor).