Enhanced situational awareness in autonomous mobile robots using context-based mapping (October 2012)

Our institution has on-going research projects which utilize autonomous mobile robots in a variety of settings. These robots navigate and interact with humans and their environment. As part of this effort a framework to integrate the navigational operation and the speech interaction to react to contextual stimuli is provided. This framework provides a system which is easy to configure and modify. The basis of this framework blends the rationale of human nature with the interpretation of sensor inputs. This combination of real-time and environmental information is at the core of having situational awareness. Context-based mapping allows the system to learn an environmental context and how to identify it from real-time interaction. Context-based mapping techniques link data prioritized by contextual value to physical locations on a visual or representative map. This system categorizes objects and events in an adaptive way. By determining the appropriate behavior in a given situation, a mobile robot uses only the relevant knowledge and data. This information is stored to allow both historical and real-time data to be used as appropriate. This approach allows the robot to access the previously collected data for statistical reference. As data are collected from various sensory inputs, the weighted contribution in that context is determined. This research examined the deployment of an autonomous mobile robot. The robot was able to function in the environment, which included both indoor and outdoor settings. Speech was used for input and as response explanations by the robot. The robot was trained initially in a supervised mode, but after the heuristics and adjustments had been reached, the robot was able to balance the sensors appropriately and learn without supervision. The result was a robot that could navigate autonomously and respond to the environment appropriately. Experiments were performed to demonstrate the ability of the robot to function effectively in indoor and outdoor environments and transition between them. The robot was also able to create a defined signature for a location using sensor information. After this training, the robot was able to exhibit situational awareness in dynamic environments, answering numerous questions regarding the state of the surrounding environment.