As for the comments above concerning motor current measurement being too slow, I offer the following considerations:
First motor lead current is the vector sum of field-producing amps and torque-producing amps. At very light torque loads, the torque amps get almost lost in the field amps which are generally between 20 and 25% of nameplate FLA. So, at light loads, using motor lead current to find torque results in large error and is usually unsuitable. However, at loads above 50% or motor rated torque, the torque amps dominate the vector equation and the result is very good accuracy. In fact, above 80% torque load, using motor lead current without solving for torque amps only results in a few % error and is generally regarded as a good indicator of motor torque (you can do the math!).
Second, motor and lead inductance are not a factor in measurement response time because it is the current that actually produces the torque (well, ok, the current produces the field and magnetic flux which then produces the torque). Inductance will retard the buildup of current which subsequently retards the buildup of torque. This actually works in our favor on this application because the motor cannot build torque instantaneously thereby giving the detection system time to respond.
Third, the slow increase in motor torque is only true for torque produced by current. When a jam occurs, the system inertia which includes the motor rotor, gearing, and the mass of the moving conveyor will produce a very rapid increase in force at the point of the jam due to rapid deceleration. I know of no method, mechanical or electrical, which will detect this at any point other than at the jam itself.
I've had experience with several similar systems including one particularly fragile expensive wire conveyor for cooling freshly baked loaves of bread. It didn't help that the whole thing was made of 316 stainless steel either. It wasn't a Mitsubishi drive but I used a similar current level detection system in the drive and, in very careful testing, routinely stopped the motor faster than any mechanical torque detector or motor lead current detector.
While this testing was done on a plain-jane V/Hz drive, if you happen to be using a sensorless vector drive, the response will be even faster. In fact, if you are using a high performance sensorless vector drive, it is likely that the drive regulator is rapidly solving for motor torque on a continuous basis and your detection can actually be done with torque rather than current. I doubt that this would make any practical different at higher motor loads, but, at light loads, the accuracy would improve dramatically.
Hope this sheds a little light on why I would prefer the drive level detector in this application. The fact that it comes almost for free doesn't hurt either!