Military Robot Laws: A Continuum of Force?

The 1999 Marine Corps Close Combat Manual (MCRP 3-02B) presents a “Continuum of Force” broken down as follows:

• Level 1: Compliant (Cooperative). The subject responds and complies to verbal commands. Close combat techniques do not apply.
• Level 2: Resistant (Passive). The subject resists verbal commands but complies immediately to any contact controls. Close combat techniques do not apply.
• Level 3: Resistant (Active). The subject initially demonstrates physical resistance. Use compliance techniques to control the situation. Level three incorporates close combat techniques to physically force a subject to comply. Techniques include: Come-along holds, Soft-handed stunning blows, Pain compliance through the use of joint manipulation and the use of pressure points.
• Level 4: Assaultive (Bodily Harm). The subject may physically attack, but does not use a weapon. Use defensive tactics to neutralize the threat. Defensive tactics include Blocks, Strikes, Kicks, Enhanced pain compliance procedures, Impact weapon blocks and blows.
• Level 5: Assaultive (Lethal Force). The subject usually has a weapon and will either kill or injure someone if he/she is not stopped immediately and brought under control. The subject must be controlled by the use of deadly force with or without a firearm.

These levels may be found here.

How should we go about programming robots for use in military situations with ethics?

Would a system like the one above, in which the level of force is determined by the level of resistance, be adequate to govern the behavior of military machines? Have your say below.