Sensing and World Modeling of Military Robotic

The mission success of any robot highly depends on world model and its sensors. The quality of sensor gathered information is important for tele-operated robots that pass this information to an operator directly, but also more for autonomous robots that use their sensor information for autonomous navigation and all sorts of autonomous robots as this is the robot’s total view on the outside world and the robot’s basis coherent execution and navigation of mission tasks.

Good world model and sensors are essential for basic information on the robot’s own location and movements, but also for tasks like route planning or automated detection, region observation, and recognition of typical targets. The right sensor to use depend on the actual tasks need to perform. For examples, current sensors are infrared sensors, CCD/HDTV sensors, acoustic and laser sensors and even radar antennas or arrays including mini SAR.

Because of the conditions variation that robots will be operated in, most sensors should be usable under all environmental conditions and weather and in all sorts of terrain. For many tasks the information should be processed on-board of the UGV, and should be efficient and effective by means of information compression and filtering of relevant information.

Concerning world modeling and sensors for military robots, following gaps were identified:
1. Obstacle negotiation and avoidance, terrain modeling and classification, and transport in normal traffic, including unstructured terrain.
2. Mine detection, de-mining, biological and chemical sensing, this gap considered not vital but important.
3. Sensor fusion at limited visibility, environmental mapping, situational awareness as well as vehicle and human detection and recognition.

The greatest challenge will be in multi sensor suites including fusion sensor, meaning that information from diverse sensors on the UGV is analyzed then merged into a more robust and complete view on the robot’s ‘outside world’ than can be achieved by any single sensor.