HistoricalHistorical ArchitecturesThese robotic architectures where an important first step, but have been surpassed by more sophisticated and appropriate architectures. However, understanding them will aid in understanding present day architectures. Sense Plan ActWho doesn't love "Shakey" the robot? Look up this group. Figure out if you can just copy over info from the TLA page. Actually... yeah, move most of that here. SubsumptionThis architecture is still of use for robots which only require a relatively low level of sophistication. The architecture demonstrated the power of its surprisingly complex emergent behavior, but unfortunately proved to be to difficult to scale to modern needs. This area of research has certainly not been abandoned. Mathematical rigor has been applied (or control theory, I suppose) to result in architectures which are able to compose behaviors in a deterministic manner. This approach is still only reasonable for simple behaviors, such as local navigation (which would need to be guided by a higher level planner to avoid local minima). But the approach works well and has been applied to groups of robots. See: Papers#Parker98ALLIANCE and Papers#Antonelli06KinematicPlatoons Also should copy a lot of stuff from the TLA page. OtherThe last two are specifically for large robot systems, I wonder if I could include any other class of systems. No need to get too broad here though, or I'm going to end up trying to describe telephone networks (or some other ancestor to sensor networks). |
