Robot Motion Planning

Usually, a real world disaster environment is partially or completely unknown and changes over time. Therefore, motion techniques must incorporate the sensory perceptions within the motion planning and the control loop. Consequently, robots have to re-plan dynamically to attain a given goal safely.

The reactive navigation module is activated by the state machine whenever a victim is detected or a goal has been provided by the path planner. It is mainly based on analyzing the structure of the environment and detecting free spaces where the robot fits in.

At the beginning of the rescue mission, no map is created and therefore the wall following module is activated. Another situation where the wall following is needed, is whenever the complete arena is explored. The advantage of the wall following is that it can easily manage to drive the robot in maze like environments.

In some situations the robot can get stuck, after sliding over a ramp and crashing against a wall. In such situations, the robot activates a recovery behavior, which tries to drive the robot towards a safe location. After that, the navigation module can take over again (e.g. exercising a wall following). The recovery behavior drives the robot by setting several consecutive motion commands based on the robot shape and the structure of the environment. See the bottom picture.