a reply to:
MykeNukem
I'm still developing the algorithms. The basic idea behind the layered intelligence is that developments that occur during operation are resolved at
the lowest level possible. For instance, each joint (freedom of direction) has a hardware controller tied to the most important sensors which operates
at a very simplistic level, but also returns data back to the next level (the leg controller). That level is software-driven, to make sure each leg
operates in a manner needed to maintain support over uneven terrain while providing the needed force to move the body. The leg controller communicates
information back to the motion controller, which monitors leg positions and adjusts to make the entire assembly move. That will then be tied back to
the master controller, which will take instructions as to the mission of the robot, combine that data with information from the motion controller,
sensory inputs, and any tools being used to determine what actions are to be taken.
In other words, let's assume the master controller determines the bot must move in the direction of a landmark that is visible. It will send the
needed signals to the leg controllers to orient the bot in the proper direction. Each leg controller will receive a signal for how it is to move
individually and send data to each joint to coordinate their movements. If anything unexpected is encountered, such as rough terrain, the joint
controller affected will sense the problem and report it to the leg controller, which will then adjust the needed parameters for the other joints to
compensate if possible. If not possible, the motion controller will instruct the legs to adjust so the issue can be handled - perhaps the body can be
lowered to the ground or raised higher. If the motion controller can't resolve the problem, it instructs the master controller to consider options to
still accomplish the goal - maybe move to the side and try again, or back up to clear an obstacle and adjust the path.
And of course, in multiple bot operations, each bot has communication ability with all other bots, including one that acts as a mission controller.
That's where the group intelligence comes in. Unlike most present assemblies, these will have the ability to change communication frequency as well,
to prevent either natural or artificial noise from jamming their signals. All communication will also be as hacker-proof as possible, with several
layers of security protocol built into all messages.
It's a pretty ambitious project, but as we venture farther and farther into space, we need bots that can solve issues and allow the mission to
proceed, even if with reduced abilities, even when one or more are damaged. Present operations rely on a single unit; if it is damaged irreparably,
the mission is over. With an army of, say, 100 smaller bots, if 50 are damaged, the mission still can proceed at 50% efficiency. The ability to simply
navigate autonomously means the speed of the mission will be increased dramatically, as the need for constant verification of all successful maneuvers
is negated.
The leg design is critical for this methodology. My leg design allows for different leg configurations to be used to accomplish the same goals, some
providing more stability, others providing more pulling power, still others providing for higher or lower ground clearance. The purpose of the
finished leg (with all 5 degrees of freedom) will be first to record various configurations to accomplish individual foot movements, so they can be
analyzed and turned into complex motion algorithms based on need. the devices will also need to have the ability to function with damaged legs... if
one is damaged, it should still be able to limp along with the remaining 5.
That's a very simplified explanation of the operation, but hopefully it will answer your question. My long term goal at this time is to develop a unit
that has all the needed capability, so the project can move into someone's hands who is more software based than I. I can't live forever, you know.
TheRedneck