That circuit has gate resistors that seem very large (100 ohms). That would slow the switch-on to perhaps 10us,
probably preventing PWM operation altogether. Suggest more like 10 ohms is a better choice to allow high speed
switching (thus reducing losses in PWM). The 18V supply will need good ceramic decoupling right at the H-bridge to
prevent severe bounce on the power rails, and more protection circuitry is really needed to reduce the risk of system failure.
Firstly the logic signals into the HIP4081 must go through resistors, perhaps 4k7's, to protect the microcontroller
when the thing fails - during testing expect problems! The HDEL and LDEL resistors need to be chosen to
ensure enough dead-time to prevent shoot-through conduction - this is vital, failure to prevent shoot-through
will usually result in MOSFETs getting extrememly hot and failing if PWM is applied.
Careful attention to PCB layout is needed in with the HIP4081 and the MOSFETs as described in the datasheet.
Go an have a search for "open source motor controller" project too.
The good news is that 18V is quite low and lower supply means fewer failure modes to worry about in an Hbridge.
The high currents will pose some issues, in particular you will need to detect overcurrent and shutdown the bridge
very quickly - I-squared-R losses in the MOSFETs are important, as are switching losses.
You must expect to test the thing first with a low current load and supply, gradually building up to higher values
if all is well - its surprisingly easy to blow up all the the FETs and the 4081 if you don't proceed very cautiously.
50A really is pushing the devices to their limit (forget what the max I specification is, calculate the power
dissipation - 10W per device (not including PWM switching losses, which may double that figure). Consider 2 FETs
per leg of the bridge, a fan-cooled heatsink is a given too.