Given a large enough tank with still water, your submersible (unless it is a perfect sphere) will execute circular motion in the steady state (after executing a number of outward spirals).
The difference in propeller RPMs results in different forces acting on the submersible at the positions of the two propellers. This results in an unbalanced torque about the center of mass.
You can represent the force on the sub (from the props) in terms of a net force acting through the CoM and a net torque about the CoM. The net force causes a linear acceleration, and the net torque results in a rotation (more accurately, an angular acceleration). As the linear and angular velocities increase, so does the viscous drag force (and torque) from the water.
In the steady state, the linear viscous drag equals the provided force, and the net force on the CoM is zero. The submersible will hence have a terminal speed determined by the linear drag coefficient (C1). Also the rotational drag will equal the supplied torque about the CoM and the angular velocity will hence remain constant thereafter, and its value will depend on coefficient of rotational drag (C2). At any point of time, (before or after the steady state is reached), the ratio of linear to angular velocities will give you the radius of curvature.
If you want your submersible to go straight, you must either change the speed of one of your two props or adjust its position a little bit. Incidentally, how are your props driven ? Are they not both running off the same motor ?