History: MicroRaptor Sensing

With MicroRaptor, I'm trying to model the autonomous controller much more like a person's brain, so doing sensory input is not as simple as in a normal robot, where you have a clear execution path.

MicroRaptor will need event-driven sensory input, in a generalized fashion. All sensory input will be handled this way, including vision and balance.

Event-driven sensory input will be handled by a sensor query engine. The sensor query engine will be instantiated into a short-term memory frame when the robot is started, and will handle all the active sensor queries.

A sensor query is basically a callback for a specific sensor condition. So, for example, if we wanted to handle balance, we could set up a sensor query that specified a particular range that the IMU has to be within. If the IMU values go outside this range, the robot is most likely falling over, and so the event handler that was registered with the sensor query is called. Sensor queries can be one-time or continuous.

That event handler might do something simple like increase the amount of force being applied to one or more servos. If the robot is off balance by more, it may do something more drastic like kick a leg out to try and prevent it from falling. In the worst case, if the fall is not preventable, it may move specific joints to try and protect the robot's head.

Landmark-based navigation can be done using sensor queries. If the robot is looking for a specific visual target, it can instantiate a visual sensor query using an object detection algorithm.