I agree with Hero; the opto-isolators as you used them do nothing but get in the way.
Here is an LTSpice sim of something that uses non-darlington transistors that will work. It can be driven with two non-overlapping logic signals which share a common ground with the 80V supply. The CW and CCW signals MUST never be asserted at the same time; there must be a few 10s of us deadband between CW and CCW, even if using the opposite direction signal for dynamic braking.
The power dissipation in the H-Bridge power transistors is small. Current spiking is reasonable. All of the transistors must have a Vce voltage rating of >150V. All of the transistors are well saturated when on, with a base current about 1/10 of their respective collector currents. They should be mounted on a heatsink, however.
This circuit does have one major issue, which is power dissipation in R4 and R7; they must be 25W resistors. If you substitute the TIP transistors for my non-darlington transistors, you should be able to reduce the current that flows in R4 and R7 by about factor of 50, reducing the power dissipation. I personally would not try to drive an 80V inductive load with 100V transistors, however...