This paper describes an approach to motion planning that includes hazards individually characterized as continuous functions of space. These functions contribute to a complex and discontinuous yet practical representation of the hazards found in a manipulator's workspace. The hazard model is a smoothed and scaled representation of the actual physical hazard sampled discretely over the robot's workspace. This research is primarily motivated to reduce damage to manipulators working in high-radiation environments, but is easily extended to other spatial hazards including heat sources, overlapping workspaces, etc. The gradient of the hazard function is used to generate a force that can be included by an artificial potential field motion planner easing its integration with other existing techniques used for obstacle avoidance, target acquisition, etc. The motion planner additionally scales the robot's velocity in proportion to the magnitude of the hazard model and determines the path. This results in a motion influenced in the direction of greatest hazard reduction at a speed that reduces the time the robot is subject to abnormally high hazard. These techniques are demonstrated on a port-deployed glovebox manipulator in a simulated hazardous environment. Over the course of a demonstration task, the radiation exposure to the robot is reduced by over 50%.